The distributional and corrective implications of alcohol ...
Transcript of The distributional and corrective implications of alcohol ...
The distributional and corrective implicationsof alcohol price policies
Rachel Griffithjoint work with Martin O’Connell and Kate Smith
DavisApril 2019
Motivation
I Large array of policies aim at reducing harm from excess alcoholconsumption
I Policies that raise prices, through taxes and regulations, are oneimportant policy
I Externalities and internalities lead to over consumption
I Policies aim to reduce this excess consumption
I where they raise the price of non-externality generatingconsumption this incurs a welfare loss
I this may be small on each individual but it can be large in aggregate
Social costs associated with alcohol
Health costs of alcohol consumption:
I e.g. roughly 70% of liver cirrhosis is attributable to alcohol
Violence and crime:
I some estimates suggest that almost half of all violent crime isalcohol related
I around 1/3 domestic violence occurs when the perpetrator isunder the influence of alcohol
I the alcohol attributable fraction of road traffic deaths is 16.6% formen and 6.7% for women
Motivation
I There are many different form of taxes and regulations
I taxes per litres/gallon and ad valorem taxes (on price) are the mostcommon
I few countries use specific taxes on alcohol content
I recently some countries have introduced minimum unit prices
I How effective are existing policies at reducing consumption thatleads to externalities/internalities?
I Will proposed reforms in Europe (and the US) make taxes bettertargetted?
I What will be the revenue implications?
How well targeted is policy?
I A policy is well targeted if it reduce purchases most by thosewhose marginal consumption creates the large externalities
I in the case of alcohol these are overwhelmingly heavy drinkers
I In order to understand how well targeted is policy we need toknow:
I something about the scale of marginal externalities, how convexthey are and how they vary across individuals
I how policy feeds through to prices
I the shape of demand and how that varies across individuals
Alcohol consumption is concentratedWe assume that marginal externalities are as well
(a) UK
Gini coefficient = 0.69
020
4060
8010
0C
umul
ativ
e sh
are
(%)
of a
lcoh
ol c
onsu
mpt
ion
0 20 40 60 80 100
Cumulative share (%) of peoplefrom lightest to heaviest drinkers
(b) US
Gini coefficient = 0.60
020
4060
8010
0C
umul
ativ
e sh
are
(%)
of a
lcoh
ol c
onsu
mpt
ion
0 20 40 60 80 100
Cumulative share (%) of peoplefrom lightest to heaviest drinkers
Source: UK Health Survey for England 2011; US National Health and Nutrition Survey 2007–2012.
Ethanol consumption, frequency and binge drinking(a) UK: frequency of drinking
02
46
8Fre
quen
cy o
f dri
nkin
g (d
ays
per
wee
k)
0 10 20 30 40Drinks per week
(b) US: frequency of drinking
12
34
5Fre
quen
cy o
f dri
nkin
g (d
ays
per
wee
k)
0 10 20 30 40Drinks per week
(c) UK: binge drinking
0.2
.4.6
.81
Pro
pens
ity
to b
inge
dri
nk
0 10 20 30 40Drinks per week
(d) US: binge drinking
0.5
11.
5Fre
quen
cy o
f bin
ge d
rink
ing
(day
s pe
r w
eek)
0 10 20 30 40Drinks per week
Source: UK Health Survey for England 2011; US National Health and Nutrition Survey 2007–2012.
Variation in the marginal externality/internality
Evidence suggests that externalities are convex in alcohol consumption
I the more you drink the greater the external cost associated withone more drink
I threshold effect with some diseases: at low levels of alcoholconsumption the risk is not elevated, but this risk increases sharplyabove a certain point
I higher levels of alcohol consumption create an exponential risk ofaccidents: odds of injury from 8 pints almost 18 times greater thanthe odds of injury from 1 pint
I It could also be that some people generate more externality (forany given level of drink) than others
Targeting and tagging of fiscal policy
I In an ideal world we would “tag” individuals and drink occasions,charging each a tax rate equal to their marginal externality(Akerlof, 1978)
I Ethanol is bundled together in products with other characteristicsand consumers have preferences over all these characteristics
I If consumer preferences over alcohol products and their overallethanol demand are correlated we can potentially use this to “tag”them and better target a tax
Heavy drinkers buy cheaper products
(a) Mean price per unit
.4.4
5.5
.55
Mea
n pr
ice
(GB
P p
er u
nit)
0 10 20 30 40 50Household's average (across year) units per adult per week
(b) Share of units bought below 45p/unit
.4.5
.6.7
.8S
hare
of u
nits
bou
ght b
elow
45p
per
uni
t
0 10 20 30 40 50Household's average (across year) units per adult per week
Unit: a unit is 10ml of ethanolSource: Kantar Worldpanel 2010-2011
Heavy drinkers buy stronger products
(a) Mean ABV of units purchased
1012
1416
18M
ean
alco
hol s
treng
th (A
BV
%)
0 10 20 30 40 50Household's average (across year) units per adult per week
(b) Share of units from segments
.51
1.5
2
Sha
re o
f uni
ts (i
ndex
ed to
1 a
tle
ss th
an 2
uni
ts p
er a
dult
per w
eek)
0 10 20 30 40 50Household's average (across year) units per adult per week
Beer Wine Spirits Cider
ABV: Alcohol By VolumeSource: Kantar Worldpanel 2010-2011
Current policies are not well targetted
I ideally we would have a specific tax on ethanol content:
pzj = cj + τ zzj
I best targeted, but not so common
I Volumetric tax:pv
j = cj + τ v vj
I Ad valorem tax:pa
j = (1 + τa)cj
j : alcohol productscj : pre-tax price of product jvj : the volume (in litres) of product jzj : units of ethanol (a unit is 10 ml of ethanol) in product j
Minimum unit prices (MUP)
I Make it illegal to sell below a price floor, p, set per unit of ethanol
I each product faces a price floor which is proportional to the numberof units of alcohol it contains
I in the absence of tax the price floor for product j is:
pmj = max
{cj ,pzj
}I In practice these policies are typically combined
I e.g a minimum unit price applied to a product that is subject to avolumetric tax plus an ad valorem tax (applied to the price inclusiveof the volumetric tax) would imply a price:
pmj = max
{(1 + τa)(cj + τ v vj),pzj
}
UK volumetric and specific taxes on alcohol
0.2
.4.6
.8Ex
cise
dut
y pe
r 10m
l of e
than
ol (G
BP)
0 10 20 30Alcohol by volume (%)
Beer Cider Wine Spirits
Note: General VAT rate also applies to alcoholic beverages
US volumetric and specific taxes on alcohol
(a) US federal and median state
010
2030
40Ex
cise
dut
y pe
r 10m
l of e
than
ol (U
SD)
0 10 20 30 40Alcoholic strength (ABV %)
Beer Wine Spirits
(b) US federal and California
010
2030
Exci
se d
uty
per 1
0ml o
f eth
anol
(USD
)
0 10 20 30 40Alcoholic strength (ABV %)
Beer Wine Spirits
Note: General and additional sales taxes also applies to alcoholic beverages and vary by State
Estimate demand
I Recall that, if consumer preferences over alcohol products iscorrelated with their overall ethanol demand, then we canpotentially use this to better target a tax
I We need estimates of individual demand for different alcoholproducts to understand whether this will be feasible
I in earlier work – Griffith, O’Connell and Smith (2018, JPubE) – weconsidered (constrained) optimal tax problem
I here we consider recent and proposed policy reforms anddistributional considerations
I we estimate a discrete choice demand model, lots of heterogeneityin preferences, this is important to capture potential correlations,and include income to study distributional consequences
Estimate demand - data on households
I Kantar Worldpanel, market research data
I alcohol purchased and brought home for consumption
I 11,634 households over 2010-2011
I everything we do is per adult member of the household
I heterogeneity in preferences
I use data over one year (2010) to establish whether adults in thehousehold are habitually light/medium/heavy drinkers
I use 2011 to estimate demand
I split by income terciles - low/moderate/high income
Distribution of alcohol purchases across householdsLight
drinkersModeratedrinkers
Heavydrinkers
0.0
5.1
.15
Fra
ctio
n of
hou
seho
lds
0 10 20 30 40Household's average (across year) units per adult per week
Source: Kantar Worldpanel 2010
... by income
Lightdrinkers
Moderatedrinkers
Heavydrinkers
0.0
5.1
.15
.2
Fra
ctio
n of
hou
seho
lds
withi
n ea
ch in
com
e gr
oup
0 10 20 30 40Household's average (across year) units per adult per week
Low income Middle incomeHigh income
Source: Kantar Worldpanel 2010
Estimate demand - data on products
There are 7000+ alcohol UPCs (barcodes) and 3000 brands
I estimating a full choice model is infeasible
I we aggregate to 70 choices
I 32 “products” available in different sizes plus an outside option tonot purchase
I focus on the margins of substitution that are most relevant to ourapplication
I we capture heterogeneity in the shape of demand for sets of UPCsthat are impacted similarly by policy reforms
Product definition and characteristics
(1) (2) (3) (4) (5) (6)Product Top brand and No. Mean Market No.definition within-product share (%) brands ABV share (%) sizes
Beer
Premium beer; ABV < 5% Newcastle Brown Ale (6.1) 386 4.4 1.8 3Premium beer; ABV ≥ 5% Old Speckled Hen (16.5) 238 5.5 2.1 3Mid-range bottled beer Budweiser Lager (19.6) 94 4.7 4.6 3Mid-range canned beer; ABV < 4.5% Carlsberg Lager (28.8) 17 3.9 5.8 3Mid-range canned beer; ABV ≥ 4.5% Stella Artois Lager (72.0) 15 5.0 2.7 3Budget beer John Smiths Bitter (23.6) 72 4.2 3.2 3
Wine
Red wine Tesco Wine (6.2) 439 12.6 18.4 4White wine Tesco Red Wine (7.8) 327 12.1 17.1 4Rose wine Echo Falls Wine (8.6) 67 11.5 4.2 2Sparkling wine Lambrini Sparkling Wine (8.4) 125 9.2 3.1 2Champagne Lanson Champagne (12.7) 42 11.8 0.8 1Port Dows Port (22.0) 23 19.8 0.7 1Sherry Harveys Bristol Cream (18.7) 25 16.8 1.2 1Vermouth Martini Extra Dry (11.8) 33 15.0 0.6 1Other fortified wines Tesco Fortified Wine (21.8) 37 14.6 0.9 1
Product definition and characteristics
(1) (2) (3) (4) (5) (6)Product Top brand and No. Mean Market No.definition within-product share (%) brands ABV share (%) sizes
Spirits
Premium gin Gordons Gin (59.6) 21 38.3 1.6 2Budget gin Tesco Gin (22.3) 15 38.3 1.3 2Premium vodka Smirnoff Red Vodka (39.0) 54 37.6 3.1 2Budget vodka Tesco Vodka (31.4) 17 37.5 1.8 2Premium whiskey Jack Daniels Bourbon/Rye (19.6) 80 40.5 2.1 2Budget whiskey Bells Scotch Whiskey (18.7) 56 40.0 8.1 2Liqueurs; ABV <30% Baileys (25.9) 203 18.4 3.1 2Liqueurs; ABV ≥30% Southern Comfort (27.2) 41 37.0 0.8 2Brandy Tesco Brandy (22.1) 55 37.3 2.4 2Rum Bacardi White Rum (29.1) 58 37.1 2.0 2Pre-mixed spirits Gordons Gin+Tonic (14.7) 43 6.1 0.2 1Alcopops Smirnoff Ice Vodka Mix (17.3) 147 4.8 0.8 1
Cider
Apple cider, <5% ABV Magners Original Cider (26.9) 52 4.4 1.6 3Apple cider, 5-6% ABV Strongbow Cider (63.1) 49 5.3 2.0 3Apple cider, >6% ABV Scrumpy Jack Cider (18.7) 71 7.0 0.8 2Pear cider Bulmers Pear Cider (24.2) 33 4.9 0.7 2Fruit cider Jacques Fruit Cider (21.4) 48 4.4 0.5 2
Product sizes and prices
(1) (2) (3) (4) (5)Product Mean Mean Mean pricedefinition Size quantity (L) price (£) per unit (£)
Beer
Premium beer; ABV < 5% 500ml 0.51 1.61 0.711-2L 1.31 3.91 0.652.5-8L 3.78 9.08 0.58
Premium beer; ABV ≥ 5% 500ml 0.52 1.84 0.611-2L 1.35 4.21 0.572.5-10L 3.70 10.30 0.51
Mid-range bottled beer 1-2L 1.43 3.50 0.532.5-4L 3.07 6.87 0.485-14L 6.78 13.40 0.43
Mid-range canned beer; ABV < 4.5% 2-5L 3.47 6.18 0.457-10L 8.39 13.23 0.4015-25L 14.93 20.50 0.39
Mid-range canned beer; ABV ≥ 4.5% 1-3L 1.96 4.55 0.464-6L 4.26 8.80 0.418-20L 9.16 16.52 0.38
Budget beer 2-4L 2.04 3.62 0.424-6L 4.57 7.51 0.428-20L 8.72 12.98 0.41
Product sizes and prices
(1) (2) (3) (4) (5)Product Mean Mean Mean pricedefinition Size quantity (L) price (£) per unit (£)
Wine
Red wine 1x750ml 0.72 4.57 0.512x750ml 1.23 7.96 0.503x750ml 1.83 11.75 0.484x750ml 3.11 18.69 0.47
White wine 1x750ml 0.72 4.45 0.512x750ml 1.26 7.90 0.503x750ml 1.81 11.28 0.494x750ml 2.90 17.10 0.47
Rose wine 1x750ml 0.71 4.23 0.522x750ml 1.81 10.40 0.49
Sparkling wine 1x750ml 0.74 5.01 0.702x750ml 2.12 9.71 0.52
Champagne 1-2x750ml 1.08 24.51 1.82Port 1-2x750ml 0.85 8.74 0.52Sherry 1-2x750ml 1.13 7.87 0.42Vermouth 1-2x750ml 1.18 6.88 0.40Other fortified wines 1-2x750ml 1.20 6.53 0.40
Product sizes and prices
(1) (2) (3) (4) (5)Product Mean Mean Mean pricedefinition Size quantity (L) price (£) per unit (£)
Spirits
Premium gin 1x700ml 0.68 11.96 0.462x700ml 1.14 17.75 0.42
Budget gin 1x700ml 0.74 9.68 0.342x700ml 1.38 12.87 0.27
Premium vodka 1x700ml 0.67 10.54 0.422x700ml 1.15 16.39 0.39
Budget vodka 1x700ml 0.57 7.73 0.362x700ml 1.14 14.01 0.34
Premium whiskey 1x700ml 0.67 19.71 0.732x700ml 1.28 29.65 0.65
Budget whiskey 1x700ml 0.65 11.02 0.422x700ml 1.17 16.18 0.40
Liqueurs; ABV <30% 1x700ml 0.64 7.84 0.682x700ml 1.26 15.94 0.66
Liqueurs; ABV ≥30% 1x700ml 0.62 13.90 0.612x700ml 1.18 19.60 0.47
Brandy 1x700ml 0.62 10.95 0.472x700ml 1.11 17.18 0.44
Rum 1x700ml 0.78 12.21 0.422x700ml 1.50 18.67 0.37
Pre-mixed spirits 700ml 0.67 4.15 0.93Alcopops 1.3L 1.30 5.95 0.87
Product sizes and prices
(1) (2) (3) (4) (5)Product Mean Mean Mean pricedefinition Size quantity (L) price (£) per unit (£)
Cider
Apple cider, <5% ABV 1L 0.91 2.56 0.622-3L 2.51 4.26 0.396-10L 6.61 10.32 0.38
Apple cider, 5-6% ABV 1-2L 1.58 2.82 0.344L 3.86 5.74 0.2810-14L 9.13 11.01 0.27
Apple cider, >6% ABV 1-2L 1.18 3.35 0.403-9L 4.19 6.88 0.27
Pear cider 1L 0.92 2.49 0.563-6L 3.86 7.61 0.42
Fruit cider 750ml 0.68 2.52 0.811-3L 1.93 6.67 0.76
What about alcohol purchased out of the home?
0.1
.2.3
.4.5
Shar
e of
alc
ohol
con
sum
ed o
ut o
f the
hom
e
Once o
r twice
a ye
ar
Once e
very co
uple
of mon
ths
Once o
r twice
a mon
th
Once o
r twice
a week
Three
or fou
r day
s a w
eek
Five
or si
x day
s a w
eek
Almost
every
day
Frequency drank any alcoholic drink in the last 12 months
Source: National Diet and Nutrition Survey
Demand model – utility
Households (i) choose which product-size (j , s) to purchase in week t ,with option to purchase no alcohol (0,0)
Utility:uijst = ν(pjsrt ,xjst ;θi) + εijst
pjsrt : pricexjst : vector of option characteristicsθi : vector of household level preference parametersεijst : idiosyncratic shock, i.i.d. type I extreme value
Utility from no purchase is normalised to:ui00t = εi00t
Demand model – utility
ν(pjsrt ,xjst ;θi)
= αipjsrt + βiwj +4∑
m=1
1[j ∈Mm] · (γim1zjs + γim2z2js) + ξijt .
pjsrt : price, product (j) - size (s) - region (r) week (t)wj : alcoholic strength (ethanol per volume)zj : ethanol – quadratic that can vary over segments (beer, wine, spiritsand cider)ξijt : household specific time varying unobserved product attribute
Demand model – preferencesI We distinguish 9 demographic groups, d = 1, . . . ,9
I light/moderate/heavy drinkers based on purchases over prior year
I low/medium/high income
I all preferences vary across these (9 groups)
I random coefficients on price, strength and ethanol – theparameters (means and standard deviations) of the distributionvarying across demographic groups
I unobserved preferences ξijt = ηij + χd(i)kt
I ηij is a time-invariant product effect, including a product effect thatvaries by demographic group and an individual specific product type(beer, wine, spirits, cider) random coefficient
I χikt is an alcohol type time effect that varies across demographicgroup
Identification of price effects
I we want to isolate causal impact of price on demand
I we absorb a lot of the demand shocks that might be correlated withprices through controls for observed and unobserved effects
I We instrument using a control function based on:
I tax rates detail
I exchange rates and producer prices detail
I retail wages index detail
I to focus on price variation driven by cost shifters
Price elasticities(a) Mean own price
-6-5
-4-3
-2M
ean
own
pric
e el
astic
ity
Light drinkers Moderate drinkers Heavy drinkers
Low income Medium income High income
(b) Mean cross price
0.0
5.1
.15
Mea
n cr
oss
pric
e el
astic
ity
Light drinkers Moderate drinkers Heavy drinkers
Low income Medium income High income
detail own detail cross
I heaviest drinkers are most responsive to prices and most likely tosubstitute towards other alcohol products - particularly lowerincome heavy drinkers
I the overall price elasticity of demand for ethanol for heavy drinkersis lower than for light drinkers
Policy reforms
I The current tax systems in the UK is poorly targeted at heavydrinkers (e.g. higher alcohol strength products are taxed less perunit of ethanol than lower strength products)
I We consider three potential reforms of the UK system
I introduction of a minimum unit price of 45p/unit, keeping specificand ad valorem taxes as they are
I reform of specific taxes to tax on ethanol at two rates, keeping advalorem taxes as they are
I increase in ad valorem tax rate from 20% to 31.5%, keeping specifictaxes as they are
I plus combinations of these policies
I We select the rates to achieve the same average reduction(around 1.15 unit per per adult per week)
Minimum unit price
I Scotland and Ireland have recently introduced minimum unitpricing of alcohol
I in Scotland prior to its introduction more than two thirds of alcoholbought in supermarkets and retailers was priced below theminimum price
I policy expected to increase prices of these products by at least35%, on average Griffith, O’Connell and Smith (2017)
I A minimum unit price raises the price of cheap alcohol, which isdisproportionately purchased by the heaviest and lowest incomedrinkers
Minimum unit price - distribution of price per unit
0.0
2.0
4.0
6.0
8S
hare
of t
rans
actio
ns
0 50 100 150Price (p/unit)
Source: Kantar Worldpanel 2010-2011
Share of units bought below 45p/unit, by incomeLight
drinkersModeratedrinkers
Heavydrinkers
.2.4
.6.8
Shar
e of
uni
ts b
ough
t be
low
45p
per
uni
t
0 10 20 30 40 50Household's average (across year) units per adult per week
Low income Middle incomeHigh income
Source: Kantar Worldpanel 2010-2011
Two-rate tax reform
010
2030
40E
xcis
e du
ty (
p pe
r un
it o
f eth
anol
)
0 10 20 30 40Alcohol by volume (%)
All alcohol
Corrective implications of policy reforms
-3-2
-10
Cha
nge
in u
nits
per
adu
lt pe
r wee
k
Light drinkers Moderate drinkers Heavy drinkers
Minimum unit price Two-rate excise tax Ad valorem increase
Aggregate implications of policy reforms
-1,0
00-5
000
500
Cha
nge
(£ m
illio
n)
Consumer surplus Industry revenue Tax revenue
Minimum unit price Two-rate excise tax Ad valorem increase
Aggregate implications of policy reforms
MUP Two rate reform
Consumer surplus -653.6 -495.0Industry revenue 213.5 -60.6Tax revenue -550.2 -618.2of which:Duty revenue -494.1 -505.1
Mechanical 0 843.3Switching -494.8 -1348.7
VAT revenue -56.1 -113.1Mechanical 148.9 168.7Switching -208.8 -281.2
Notes: £million per year.
Effects of policy reforms
I How well targeted?
I Minimum unit price and a two-rate reform achieve similar reductionsin units purchased across light, moderate and heavy drinkers
I an increase in ad valorem taxes is less well targeted, leading tohigher reduction amongst light and lower amongst heavy drinkers
I What are the revenue implications?
I Minimum unit price and two rate reform both reduce tax revenuesrelative to the current UK tax system, reflecting their better successat discouraging drinking
I Under minimum unit price some of that is shifted to industry, andconsumer surplus declines by more compared with two-rate reform
I an increase in ad valorem taxes raises some revenue but leads tolarge reductions in consumer surplus and industry revenue
Concerns about regressivity of corrective taxes
I A common concern about excise duties is that they are regressive
I Ultimately what matters for meeting distributive goals is thedistributional impacts of the tax and benefit system as a whole,not the progressivity or regressivity of any single tax:
I excise taxes are an effective way of tackling externalities as theyalter the relative prices consumers face
I policy makers can offset the regressivity of corrective taxes throughadjustments to other parts of the system
I understanding the distributional impact of excise taxes is importantto understand how to do this
Distributional implications of policy reforms
-150
-100
-50
0C
hang
e in
con
sum
er s
urpl
us (£
per
hh
per y
ear)
Low income Middle income High income
Minimum unit price Two-rate excise tax Ad valorem increase
Summary
I Minimum unit prices and corrective taxes are both effectiveinstruments for reducing externalities in alcohol markets
I Miniumum unit prices have the disadvantage that they transferrevenue to industry and they are regressive
I Ad valorem taxes are badly targeted and increasing them (fromcurrent UK levels) will raise only small amounts of revenue
I These policies can be effectively combined, and reforms to taxesin combination with minimum unit prices might be the best policy
I To better understand optimal policy we would need to have bettermeasures of externalities at the individual level
Instruments - tax changes in 2011 Back
Segment Applies to products: Rate changes (month)
Beer 1.8-2.8% ABV +1.25 (March); -9.28 (Oct)2.8-7.5% ABV +1.25 (March)>7.5% ABV +1.25 (March); +4.64 (Oct)
Wine 5.5-15% ABV (still) +16.23 (March)15-22% ABV (still) +21.64 (March)5.5-8.5% ABV (sparkling) +15.72 (March)8.5-15% ABV (sparkling) +20.79 (March)
Spirits 0-100% ABV +1.72 (March)Cider 1.2-7.5% ABV -0.14 (March)
7.5-8.5% ABV -0.17 (March)
Instruments - exchange rates and producer prices Back
(a) Exchange rates
1.56
1.58
1.6
1.62
1.64
USD
-GBP
1.12
1.14
1.16
1.18
EUR-
GBP
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
EUR-GBP exchange rate USD-GBP exchange rate
(b) Factory gate prices for beer andcider
100
102
104
106
108
110
Fact
ory
gate
pric
e in
dex
(Jan
201
0 =
100)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Factory gate price of beer Factory gate price of cider
Instruments - local economic conditions (wages) Back
Retailer coverage in different regions
Tesco Sainsbury’s Asda Morrisons Discounter Upmarket Other
North East 19.3 10.0 27.6 18.7 5.9 2.6 16.0North West 27.1 9.8 25.3 14.6 6.0 2.2 15.0Yorkshire and Humber 23.3 9.9 21.3 23.3 4.8 2.5 14.9East Midlands 29.9 13.2 17.8 15.8 5.4 2.6 15.3West Midlands 26.0 15.6 19.6 14.8 6.2 2.8 14.9East of England 39.3 16.5 13.6 9.6 4.2 4.4 12.4London 31.7 26.2 11.9 7.2 3.3 6.4 13.2South East 36.0 22.1 14.1 7.5 3.8 5.6 11.0South West 34.5 17.4 14.9 11.4 6.0 4.2 11.5Wales 34.0 6.8 22.1 12.1 7.7 2.3 14.9Scotland 30.8 7.4 22.5 15.8 6.5 2.8 14.1
Combinations of MUP and tax reform
Two-rate taxes (p/unit)
MUP (p) <20% ABV ≥20% ABV
30.0 18.8 37.532.5 18.8 37.535.0 18.7 37.437.5 18.6 37.140.0 18.2 36.342.5 17.4 34.845.0 16.2 32.4
Effect on units purchased by heavy relative to lightdrinkers
67
89
10
Red
uctio
n in
uni
ts p
urch
ased
by
heav
y dr
inke
rs/
redu
ctio
n in
uni
ts p
urch
ased
by
light
drin
kers
30 35 40 45Value of MUP
(tax rates adjusted so aggregate decline in units held constant)
MUP + two-rate excise tax Two-rate excise tax only