SlideShare a Scribd company logo
Hedonic Models and New Characteristics
Ian Crawford
Oxford and IFS
J. Peter Neary
Oxford, CEPR and CESifo
Challenges and Opportunities in Price Measurement
The Shard
July 19, 2018
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 1 / 41
Introduction
Outline
How to correct measured inflation for quality change?
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
But the standard method does not account for changes to product
specifications on the extensive margin. This can lead to bias.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
But the standard method does not account for changes to product
specifications on the extensive margin. This can lead to bias.
We show how to correct for it and provide an empirical example.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
But the standard method does not account for changes to product
specifications on the extensive margin. This can lead to bias.
We show how to correct for it and provide an empirical example.
Our methodological innovation is to combine:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
But the standard method does not account for changes to product
specifications on the extensive margin. This can lead to bias.
We show how to correct for it and provide an empirical example.
Our methodological innovation is to combine:
Linear characteristics model of Gorman (1956)
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
How to correct measured inflation for quality change?
National statistical agencies, like the ONS or the BLS or Eurostat,
now routinely use “hedonic” methods to correct their measures of
inflation.
But the standard method does not account for changes to product
specifications on the extensive margin. This can lead to bias.
We show how to correct for it and provide an empirical example.
Our methodological innovation is to combine:
Linear characteristics model of Gorman (1956)
Sato (1976) - Vartia (1976) - Feenstra (1994) index number for new
goods
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
Introduction
Outline
1 Background Theory
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 3 / 41
Background Theory
Outline
1 Background Theory
Measuring Inflation
Hedonic Methods
The Linear Characteristics Model
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 4 / 41
Background Theory Measuring Inflation
The True Cost-of-Living Index
A utility-maximizing consumer:
x1
t , ..., xK
t = arg max
x1,...,xK
u(x1
, ..., xK
) ∑
k
pk
t xk
= yt
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
Background Theory Measuring Inflation
The True Cost-of-Living Index
A utility-maximizing consumer:
x1
t , ..., xK
t = arg max
x1,...,xK
u(x1
, ..., xK
) ∑
k
pk
t xk
= yt
The expenditure function:
c(pt, u) = min
x1,...,xK
∑
j
p
j
txj
u(x1
, ..., xK
) = u
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
Background Theory Measuring Inflation
The True Cost-of-Living Index
A utility-maximizing consumer:
x1
t , ..., xK
t = arg max
x1,...,xK
u(x1
, ..., xK
) ∑
k
pk
t xk
= yt
The expenditure function:
c(pt, u) = min
x1,...,xK
∑
j
p
j
txj
u(x1
, ..., xK
) = u
The constant-utility Kon¨us price index is defined as:
PK(p0, p1, u) =
c(p1, u)
c(p0, u)
Characteristics New Characteristics
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
Background Theory Measuring Inflation
Observable Price Indexes
Standard index number theory then provides a number of ways of
either approximating or, subject to functional form assumptions,
exactly computing the true index.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
Background Theory Measuring Inflation
Observable Price Indexes
Standard index number theory then provides a number of ways of
either approximating or, subject to functional form assumptions,
exactly computing the true index.
For example:
PL = ∑
j
p
j
1
p
j
0
s
j
0 PP =

∑
j
p
j
1
p
j
0
−1
s
j
1


−1
PF = PLPP PT = ∏
j
p
j
1
p
j
0
1
2 (s
j
0+s
j
1)
where s
j
t =
p
j
tx
j
t
∑i pi
txi
t
.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
Background Theory Measuring Inflation
Observable Price Indexes
Standard index number theory then provides a number of ways of
either approximating or, subject to functional form assumptions,
exactly computing the true index.
For example:
PL = ∑
j
p
j
1
p
j
0
s
j
0 PP =

∑
j
p
j
1
p
j
0
−1
s
j
1


−1
PF = PLPP PT = ∏
j
p
j
1
p
j
0
1
2 (s
j
0+s
j
1)
where s
j
t =
p
j
tx
j
t
∑i pi
txi
t
.
Problem: No allowance for quality change
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
Background Theory Hedonic Methods
Hedonic Methods
There are two basic methods of correcting for quality change:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
Background Theory Hedonic Methods
Hedonic Methods
There are two basic methods of correcting for quality change:
1 the time dummy method - based on the repackaging model of Fisher
and Shell (1971) and Muellbauer (1974)
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
Background Theory Hedonic Methods
Hedonic Methods
There are two basic methods of correcting for quality change:
1 the time dummy method - based on the repackaging model of Fisher
and Shell (1971) and Muellbauer (1974)
2 the indirect method - based on the linear characteristics model of
Gorman (1956) and Lancaster (1966).
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
Background Theory Hedonic Methods
The Linear Characteristics Model
There are K varieties of products with quantities denoted by xk and
unit prices by pk.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
Background Theory Hedonic Methods
The Linear Characteristics Model
There are K varieties of products with quantities denoted by xk and
unit prices by pk.
These products are differentiated by J characteristics.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
Background Theory Hedonic Methods
The Linear Characteristics Model
There are K varieties of products with quantities denoted by xk and
unit prices by pk.
These products are differentiated by J characteristics.
In the linear characteristics model the total amount of a given
characteristic present in a bundle of varieties x1
t , ..., xK
t observed in
period t is
z
j
t = ∑
k
a
k,j
t xk
t
where a
k,j
t represents the amount of characteristic j present in one
unit of product k according to the product specification in period t.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
The consumer’s problem is:
max
x1,...,xK
v(z1
, ..., zJ
)
subject to: ∑
k
pk
t xk
= yt and zj
= ∑
k
a
k,j
t xk
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
The consumer’s problem is:
max
x1,...,xK
v(z1
, ..., zJ
)
subject to: ∑
k
pk
t xk
= yt and zj
= ∑
k
a
k,j
t xk
Maximising behaviour implies the first order conditions:
pk
t = ∑
j
a
k,j
t π
j
t ∀k, t
where π
j
t ≡ 1
ψt
v
j
t are the shadow prices of characteristics.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
The consumer’s problem is:
max
x1,...,xK
v(z1
, ..., zJ
)
subject to: ∑
k
pk
t xk
= yt and zj
= ∑
k
a
k,j
t xk
Maximising behaviour implies the first order conditions:
pk
t = ∑
j
a
k,j
t π
j
t ∀k, t
where π
j
t ≡ 1
ψt
v
j
t are the shadow prices of characteristics.
This is the hedonic pricing equation.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
The budget constraint “adds up” in both characteristics and product
space:
∑
k
pk
t xk
t = ∑
j
π
j
tz
j
t = yt
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 10 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
The budget constraint “adds up” in both characteristics and product
space:
∑
k
pk
t xk
t = ∑
j
π
j
tz
j
t = yt
The model can be thought of in terms of either preferences-
for-products or preferences-for-characteristics:
x1
t , ..., xK
t = arg max u(x1
, ..., xK
) ∑
k
pk
t xk
= yt
z1
t , ..., zJ
t = arg max v(z1
, ..., zJ
) ∑
j
π
j
tzj
= yt
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 10 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
z1
t , ..., zJ
t = arg max v(z1
, ..., zJ
) ∑
j
π
j
tzj
= yt
The hedonic expenditure function:
c(πt, v) = min
z1,...,zJ
∑
j
π
j
tz
j
t v(z1
, ..., zJ
) = v
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 11 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
z1
t , ..., zJ
t = arg max v(z1
, ..., zJ
) ∑
j
π
j
tzj
= yt
The hedonic expenditure function:
c(πt, v) = min
z1,...,zJ
∑
j
π
j
tz
j
t v(z1
, ..., zJ
) = v
The constant-utility Kon¨us hedonic price index is defined as:
PK(π0, π1, v) =
c(π1, v)
c(π0, v)
CostOfLiving New Characteristics
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 11 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
All of these hedonic price indices depend on shadow price-relatives:
π
j
1
π
j
0
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
All of these hedonic price indices depend on shadow price-relatives:
π
j
1
π
j
0
If a new feature is added to products in this category in period 1:
what is the denominator π
j
0?
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
Background Theory The Linear Characteristics Model
The Linear Characteristics Model
All of these hedonic price indices depend on shadow price-relatives:
π
j
1
π
j
0
If a new feature is added to products in this category in period 1:
what is the denominator π
j
0?
If an existing feature is withdrawn from the products in this category
in period 1: what is the numerator π
j
1?
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
Adjusting for New Characteristics: General
Outline
1 Background Theory
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 13 / 41
Adjusting for New Characteristics: General
Rationing Theory
The solution comes from the theory of consumer behaviour under
rationing
(Hicks (1940), Rothbarth (1941), Neary and Roberts (1980))
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
Adjusting for New Characteristics: General
Rationing Theory
The solution comes from the theory of consumer behaviour under
rationing
(Hicks (1940), Rothbarth (1941), Neary and Roberts (1980))
Central idea: Use “virtual prices” ˜π
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
Adjusting for New Characteristics: General
Rationing Theory
The solution comes from the theory of consumer behaviour under
rationing
(Hicks (1940), Rothbarth (1941), Neary and Roberts (1980))
Central idea: Use “virtual prices” ˜π
Prices that would induce the consumer to consume voluntarily the
constrained characteristic
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
Adjusting for New Characteristics: General
Rationing Theory
The solution comes from the theory of consumer behaviour under
rationing
(Hicks (1940), Rothbarth (1941), Neary and Roberts (1980))
Central idea: Use “virtual prices” ˜π
Prices that would induce the consumer to consume voluntarily the
constrained characteristic
Here, the “constraint” is that a characteristic is unavailable in one
period
Skip Details
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
Adjusting for New Characteristics: General
Rationing Theory
Consider the constrained hedonic cost function
˜c(πt, v, ¯z1
t ) = min
z2,...,zJ
π1
t ¯z1
t + ∑
j=2
π
j
tzj
subject to v(¯z1
, ..., zJ
) = v
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 15 / 41
Adjusting for New Characteristics: General
Rationing Theory
Consider the constrained hedonic cost function
˜c(πt, v, ¯z1
t ) = min
z2,...,zJ
π1
t ¯z1
t + ∑
j=2
π
j
tzj
subject to v(¯z1
, ..., zJ
) = v
It satisfies Sheppard’s Lemma
π1 ˜c(πt, v, ¯z1
t ) = ¯z1
t
πj ˜c(πt, v, ¯z1
t ) = zj
(πt, v, ¯z1
t ) ∀j = 2, ..J
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 15 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The relationship between the constrained and unconstrained hedonic
cost functions is given by
˜c(πt, v, ¯z1
t ) = π1
t ¯z1
t + ∑
j=2
π
j
tzj
(πt, v, ¯z1
t )
= π1
t z1
( ˜πt, v) + ∑
j=2
π
j
tzj
( ˜πt, v)
= c( ˜πt, v) + π1
t − ˜π1
t ¯z1
t
where ˜π1
t is the virtual price of the constrained characteristic and
˜πt = ˜π1
t , π2
t , ..., πJ
t is the shadow support price vector.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 16 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The shadow support prices are those which would precisely induce the
consumer to choose the constrained characteristics vector
¯z1
t = z1
( ˜πt, v)
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The shadow support prices are those which would precisely induce the
consumer to choose the constrained characteristics vector
¯z1
t = z1
( ˜πt, v)
The welfare cost of the constraint on the characteristic is
∂˜c(πt, v, ¯z1
t )
∂¯z1
t
=
∂c( ˜πt, v)
∂π1
t
∂π1
t (πt, v, ¯z1
t )
∂¯z1
t
−
∂π1
t (πt, v, ¯z1
t )
∂¯z1
t
¯z1
t + π1
t − ˜π1
t
= π1
t − ˜π1
t
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The shadow support prices are those which would precisely induce the
consumer to choose the constrained characteristics vector
¯z1
t = z1
( ˜πt, v)
The welfare cost of the constraint on the characteristic is
∂˜c(πt, v, ¯z1
t )
∂¯z1
t
=
∂c( ˜πt, v)
∂π1
t
∂π1
t (πt, v, ¯z1
t )
∂¯z1
t
−
∂π1
t (πt, v, ¯z1
t )
∂¯z1
t
¯z1
t + π1
t − ˜π1
t
= π1
t − ˜π1
t
Thus the change in welfare associated with a change in the constraint
is the difference between the shadow price of the characteristic and its
virtual counterpart.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
Consider two periods t ∈ {0, 1} and suppose that the first
characteristic is not available in the base period ¯z1
0 = 0 but is
added to the product specification in the following period.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
Consider two periods t ∈ {0, 1} and suppose that the first
characteristic is not available in the base period ¯z1
0 = 0 but is
added to the product specification in the following period.
Then
˜c(π0, v, ¯z1
0) = c( ˜π0, v) + π1
0 − ˜π1
t ¯z1
0
which becomes
˜c(π0, v, 0) = c( ˜π0, v).
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
Consider two periods t ∈ {0, 1} and suppose that the first
characteristic is not available in the base period ¯z1
0 = 0 but is
added to the product specification in the following period.
Then
˜c(π0, v, ¯z1
0) = c( ˜π0, v) + π1
0 − ˜π1
t ¯z1
0
which becomes
˜c(π0, v, 0) = c( ˜π0, v).
Consequently, the Kon¨us hedonic cost-of-living index can be expressed
conveniently as the ratio of unconstrained hedonic cost functions
where the base prices are given by the virtual prices:
PK( ˜π0, π1, v) =
c(π1, v)
c( ˜π0, v)
CostOfLiving Characteristics Sato-Vartia
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
But all of them require that the shadow price relative π1
1/ ˜π1
0 is
known.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
But all of them require that the shadow price relative π1
1/ ˜π1
0 is
known.
Solutions:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
But all of them require that the shadow price relative π1
1/ ˜π1
0 is
known.
Solutions:
Ignore the affected characteristics and chain.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
But all of them require that the shadow price relative π1
1/ ˜π1
0 is
known.
Solutions:
Ignore the affected characteristics and chain.
Estimate the model and solve 0 = z1(π, v) for ˜π1.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: General
New Characteristics Bias
The fact that everything can be expressed in terms of the
unconstrained cost function means that all of the standard index
number results are available.
But all of them require that the shadow price relative π1
1/ ˜π1
0 is
known.
Solutions:
Ignore the affected characteristics and chain.
Estimate the model and solve 0 = z1(π, v) for ˜π1.
Something else ...
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
Adjusting for New Characteristics: A Structural Approach
Outline
1 Background Theory
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 20 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
To make progress, we need to impose more structure
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
To make progress, we need to impose more structure
We assume CES preferences over characteristics:
v(z) = ∑ βjz
σ−1
σ
j
σ
σ−1
= ∑ βjzθ
j
1
θ
c(π, u) = c(π)u = ∑ βσ
j π1−σ
j
1
1−σ
u
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
To make progress, we need to impose more structure
We assume CES preferences over characteristics:
v(z) = ∑ βjz
σ−1
σ
j
σ
σ−1
= ∑ βjzθ
j
1
θ
c(π, u) = c(π)u = ∑ βσ
j π1−σ
j
1
1−σ
u
where σ ≡ 1
1−θ and θ ≡ σ−1
σ .
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
The Sato-Vartia cost-of-living index is exact for CES preferences
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
The Sato-Vartia cost-of-living index is exact for CES preferences
This holds for preferences over characteristics as well as over goods:
PSV =
c(π1)
c(π0)
= ∏
j
πj,1
πj,0
ωj
where:
ωj =
µj
∑ µj
µj =
sj,1 − sj,0
ln sj,1 − ln sj,0
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia Index
The Sato-Vartia cost-of-living index is exact for CES preferences
This holds for preferences over characteristics as well as over goods:
PSV =
c(π1)
c(π0)
= ∏
j
πj,1
πj,0
ωj
where:
ωj =
µj
∑ µj
µj =
sj,1 − sj,0
ln sj,1 − ln sj,0
Note that this corresponds exactly to the true hedonic index but does
not depend on unobserved coefficients.
New Characteristics Sato-Vartia-Feenstra
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
Now consider two periods t ∈ {0, 1} and suppose that the set of
characteristics changes from J0 in the base period to J1 in the
comparison period.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
Now consider two periods t ∈ {0, 1} and suppose that the set of
characteristics changes from J0 in the base period to J1 in the
comparison period.
The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics
which are available in both periods.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
Now consider two periods t ∈ {0, 1} and suppose that the set of
characteristics changes from J0 in the base period to J1 in the
comparison period.
The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics
which are available in both periods.
We can show that:
PSVF =
c( ˜π1)
c( ˜π0)
=
λ1
λ0
1
σ−1
∏
j∈J
π
j
1
π
j
0
ωj
is the exact price index.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
Now consider two periods t ∈ {0, 1} and suppose that the set of
characteristics changes from J0 in the base period to J1 in the
comparison period.
The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics
which are available in both periods.
We can show that:
PSVF =
c( ˜π1)
c( ˜π0)
=
λ1
λ0
1
σ−1
∏
j∈J
π
j
1
π
j
0
ωj
is the exact price index.
It equals the Sato-Vartia index, times an adjustment factor
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
Now consider two periods t ∈ {0, 1} and suppose that the set of
characteristics changes from J0 in the base period to J1 in the
comparison period.
The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics
which are available in both periods.
We can show that:
PSVF =
c( ˜π1)
c( ˜π0)
=
λ1
λ0
1
σ−1
∏
j∈J
π
j
1
π
j
0
ωj
is the exact price index.
It equals the Sato-Vartia index, times an adjustment factor
This needs an estimate of σ, but otherwise depends only on
observables.
Sato-Vartia
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
The adjustment factor
λ1
λ0
1
σ−1
where
λt =
∑j∈J π
j
tz
j
t
∑j∈Jt
π
j
tz
j
t
≤ 1
takes into account the changing set of characteristics on the extensive
margin.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
The adjustment factor
λ1
λ0
1
σ−1
where
λt =
∑j∈J π
j
tz
j
t
∑j∈Jt
π
j
tz
j
t
≤ 1
takes into account the changing set of characteristics on the extensive
margin.
If new characteristics are important, λ1 will tend to be small so the
hedonic price index will be lower.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
Adjusting for New Characteristics: A Structural Approach
The Sato-Vartia-Feenstra Index
The adjustment factor
λ1
λ0
1
σ−1
where
λt =
∑j∈J π
j
tz
j
t
∑j∈Jt
π
j
tz
j
t
≤ 1
takes into account the changing set of characteristics on the extensive
margin.
If new characteristics are important, λ1 will tend to be small so the
hedonic price index will be lower.
The adjustment factor is less important the higher is the substitution
parameter σ, i.e., ignoring new characteristics matters less if they are
close substitutes for existing ones.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
An Application: New Cars, New Characteristics
Outline
1 Background Theory
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 25 / 41
An Application: New Cars, New Characteristics
The Data
Data were collected by the UK Office for National Statistics from
What Car on the major new car models available from 1988 to mid
1995.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
An Application: New Cars, New Characteristics
The Data
Data were collected by the UK Office for National Statistics from
What Car on the major new car models available from 1988 to mid
1995.
Data on market shares and sales are from the Competition
Commission.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
An Application: New Cars, New Characteristics
The Data
Data were collected by the UK Office for National Statistics from
What Car on the major new car models available from 1988 to mid
1995.
Data on market shares and sales are from the Competition
Commission.
The data covers 36 main models.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
An Application: New Cars, New Characteristics
The Data
Data were collected by the UK Office for National Statistics from
What Car on the major new car models available from 1988 to mid
1995.
Data on market shares and sales are from the Competition
Commission.
The data covers 36 main models.
Prices and specifications are measured quarterly (30 periods, 3160
observations)
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
An Application: New Cars, New Characteristics
Summary Statistics
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 27 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
and at the beginning of the second quarter, which normally coincides
with changes in vehicle excise duty and taxes on motor fuels.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
and at the beginning of the second quarter, which normally coincides
with changes in vehicle excise duty and taxes on motor fuels.
Looking at the intensive margin first, there were quality improvements
in terms of performance across the board:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
and at the beginning of the second quarter, which normally coincides
with changes in vehicle excise duty and taxes on motor fuels.
Looking at the intensive margin first, there were quality improvements
in terms of performance across the board:
0-60 mph acceleration times improved over the period by 6.5%
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
and at the beginning of the second quarter, which normally coincides
with changes in vehicle excise duty and taxes on motor fuels.
Looking at the intensive margin first, there were quality improvements
in terms of performance across the board:
0-60 mph acceleration times improved over the period by 6.5%
fuel economy improved by 2.8%.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Specifications generally change twice per year:
at the beginning of the fourth quarter, which marks the time when new
annual registration letters were issued in the UK;
and at the beginning of the second quarter, which normally coincides
with changes in vehicle excise duty and taxes on motor fuels.
Looking at the intensive margin first, there were quality improvements
in terms of performance across the board:
0-60 mph acceleration times improved over the period by 6.5%
fuel economy improved by 2.8%.
both torque and brake-horse-power improved by 5.2% on average
(torque and bhp are proportional at a fixed engine speed).
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
central locking was present in a quarter of cars at the start of the
period; this rose to nearly two-thirds by the end,
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
central locking was present in a quarter of cars at the start of the
period; this rose to nearly two-thirds by the end,
manual sun-roofs went from being available on 17.6% of models to 25%
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
central locking was present in a quarter of cars at the start of the
period; this rose to nearly two-thirds by the end,
manual sun-roofs went from being available on 17.6% of models to 25%
electric front-windows were uncommon at beginning with 8% market
share, but by the end 40% of new models had them.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
central locking was present in a quarter of cars at the start of the
period; this rose to nearly two-thirds by the end,
manual sun-roofs went from being available on 17.6% of models to 25%
electric front-windows were uncommon at beginning with 8% market
share, but by the end 40% of new models had them.
electrically operated door mirrors went from 8% to 25%
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
The availability of features also improved:
central locking was present in a quarter of cars at the start of the
period; this rose to nearly two-thirds by the end,
manual sun-roofs went from being available on 17.6% of models to 25%
electric front-windows were uncommon at beginning with 8% market
share, but by the end 40% of new models had them.
electrically operated door mirrors went from 8% to 25%
split-fold rear seats increased from being specified in around half of
new cars to nearly three-quarters.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
On the extensive margin there were a number of changes to the set of
characteristics over the period.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
On the extensive margin there were a number of changes to the set of
characteristics over the period.
electronic seat and steering-adjustability both became available in the
fourth quarter of 1988;
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
On the extensive margin there were a number of changes to the set of
characteristics over the period.
electronic seat and steering-adjustability both became available in the
fourth quarter of 1988;
electric sun-roofs were introduced in the first quarter of 1990;
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
On the extensive margin there were a number of changes to the set of
characteristics over the period.
electronic seat and steering-adjustability both became available in the
fourth quarter of 1988;
electric sun-roofs were introduced in the first quarter of 1990;
air conditioning was introduced in the fourth quarter of 1993 as were
powered (heated) seats and drivers’ airbags.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
On the extensive margin there were a number of changes to the set of
characteristics over the period.
electronic seat and steering-adjustability both became available in the
fourth quarter of 1988;
electric sun-roofs were introduced in the first quarter of 1990;
air conditioning was introduced in the fourth quarter of 1993 as were
powered (heated) seats and drivers’ airbags.
CD-players were introduced in the second quarter of 1994. By then
simple radio-only models had gone - they disappeared from the new car
market by the fourth quarter of 1992.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Other features were short-lived and simply came and went:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Other features were short-lived and simply came and went:
head-lamp cleaners were available in some cars (Volvos) at the start of
our period of observation, but disappeared at the start of 1990 only to
reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and
Volvo’s again) before disappearing again.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Other features were short-lived and simply came and went:
head-lamp cleaners were available in some cars (Volvos) at the start of
our period of observation, but disappeared at the start of 1990 only to
reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and
Volvo’s again) before disappearing again.
Similarly early trip-computers became available (also on Alfa’s) but
disappeared only to be offered briefly by Rover before again
disappearing.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
An Application: New Cars, New Characteristics
Changes in Characteristics
Other features were short-lived and simply came and went:
head-lamp cleaners were available in some cars (Volvos) at the start of
our period of observation, but disappeared at the start of 1990 only to
reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and
Volvo’s again) before disappearing again.
Similarly early trip-computers became available (also on Alfa’s) but
disappeared only to be offered briefly by Rover before again
disappearing.
Trip computers are now completely standard but they didn’t take off
until after the end of our period of observation.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
An Application: New Cars, New Characteristics
The Profile of New Car Prices
The density of the prices of new cars over time
5 10 15 20
0.000.050.100.150.20
£ '000's
Density
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 32 / 41
An Application: New Cars, New Characteristics
The Profile of New Car Prices
The density of the prices of new cars over time
88 90 92 94
05101520
£'000's
88 89 90 91 92 93 94 95
200
200
200
200 200
400
400
400
400
400
400
600
600
600
600
600
800
800
1000
1000
1000
1200
1200
12001400
1400
1600
1600
1600
1600
1800
1800
1800
2000
2000
2000
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 33 / 41
An Application: New Cars, New Characteristics
Recap on Methodology
1 Estimate within period linear characteristics models for each period
using only extant characteristics to recover
{πt, zt}t=0,...,T
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
An Application: New Cars, New Characteristics
Recap on Methodology
1 Estimate within period linear characteristics models for each period
using only extant characteristics to recover
{πt, zt}t=0,...,T
2 Fit a value for σ.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
An Application: New Cars, New Characteristics
Recap on Methodology
1 Estimate within period linear characteristics models for each period
using only extant characteristics to recover
{πt, zt}t=0,...,T
2 Fit a value for σ.
3 Compute the adjustment factors
{λt}t=0,...,T
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
An Application: New Cars, New Characteristics
Recap on Methodology
1 Estimate within period linear characteristics models for each period
using only extant characteristics to recover
{πt, zt}t=0,...,T
2 Fit a value for σ.
3 Compute the adjustment factors
{λt}t=0,...,T
4 Compute the variety-adjusted Sato-Vartia hedonic price index
c(πt)
c(π0)
=
λt
λ0
1
σ−1
∏
j∈J
πj,t
πj,0
ωj
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
An Application: New Cars, New Characteristics
Empirical Results: Price Levels
88 90 92 94
0.81.01.21.4
Index
88 89 90 91 92 93 94 95
Laspeyres
Paasche
Fisher
Sato−Vartia
Sato−Vartia−Feenstra
Repackaging
Notes: 1988q4: adjustable seats and adjustable steering introduced; 1990q1: electric sun roofs introduced; 1990q2: headlamp
cleaners withdrawn; 1991q1: trip computers introduced; 1992q1: trip computers withdrawn; 1992q3: trip computers
re-introduced for one quarter; 1992q4 radio-only withdrawn; 1993q2: headlamp cleaners re-introduced; 1993q4: air conditioning,
power seats and driver’s airbags introduced; 1994q2 CD players available; 1993q4: headlamp cleaners finally withdrawn.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 35 / 41
An Application: New Cars, New Characteristics
Empirical Results: Inflation Rates
89 90 91 92 93 94 95
−0.3−0.2−0.10.00.1
Index
89 90 91 92 93 94 95
Laspeyres
Paasche
Fisher
Sato−Vartia
Sato−Vartia−Feenstra
Repackaging
Notes: 1988q4: adjustable seats and adjustable steering introduced; 1990q1: electric sun roofs introduced; 1990q2: headlamp
cleaners withdrawn; 1991q1: trip computers introduced; 1992q1: trip computers withdrawn; 1992q3: trip computers
re-introduced for one quarter; 1992q4 radio-only withdrawn; 1993q2: headlamp cleaners re-introduced; 1993q4: air conditioning,
power seats and driver’s airbags introduced; 1994q2 CD players available; 1993q4: headlamp cleaners finally withdrawn.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 36 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Hedonic quality adjustment on the basis of overlapping characteristics
reduces this to an annual average growth rate of 3.45%.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Hedonic quality adjustment on the basis of overlapping characteristics
reduces this to an annual average growth rate of 3.45%.
Allowing for the changes on the extensive margin of characteristics
gives an annual average growth rate of about 2%.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Hedonic quality adjustment on the basis of overlapping characteristics
reduces this to an annual average growth rate of 3.45%.
Allowing for the changes on the extensive margin of characteristics
gives an annual average growth rate of about 2%.
Very roughly:
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Hedonic quality adjustment on the basis of overlapping characteristics
reduces this to an annual average growth rate of 3.45%.
Allowing for the changes on the extensive margin of characteristics
gives an annual average growth rate of about 2%.
Very roughly:
if we take the Boskin Commission’s estimate of the bias from quality
change of 0.6 percentage points per year,
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
An Application: New Cars, New Characteristics
Conclusions
Without quality adjustment we see an annual average growth in prices
of new cars of about 5.37%.
Hedonic quality adjustment on the basis of overlapping characteristics
reduces this to an annual average growth rate of 3.45%.
Allowing for the changes on the extensive margin of characteristics
gives an annual average growth rate of about 2%.
Very roughly:
if we take the Boskin Commission’s estimate of the bias from quality
change of 0.6 percentage points per year,
our results suggest that allowing for changes in the extensive margin
could add half as much again to that.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
Summary and Conclusion
Outline
1 Background Theory
2 Adjusting for New Characteristics: General
3 Adjusting for New Characteristics: A Structural Approach
4 An Application: New Cars, New Characteristics
5 Summary and Conclusion
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 38 / 41
Summary and Conclusion
Summary
We introduce a new approach to correcting inflation for quality
change at the extensive margin
The theory builds on and extends:
The Gorman approach to preferences over characteristics
The Feenstra approach to correcting price indices for new goods
The method we propose is easy to apply in practice
An application to UK new car prices shows that accounting for
changes in characteristics matters:
It reduces inflation by 1.45% per annum
Half as much as accounting for quality change at the intensive margin
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 39 / 41
Summary and Conclusion
Practical Pros and Cons
The main drawback of our approach is that it is based on an
assumption.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
Summary and Conclusion
Practical Pros and Cons
The main drawback of our approach is that it is based on an
assumption.
The generalised CES is “not-quite-superlative”, and therefore we
cannot just assume that the approximation to true preferences will be
good in the vicinity of the data (although see Hill (2006)).
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
Summary and Conclusion
Practical Pros and Cons
The main drawback of our approach is that it is based on an
assumption.
The generalised CES is “not-quite-superlative”, and therefore we
cannot just assume that the approximation to true preferences will be
good in the vicinity of the data (although see Hill (2006)).
Nonetheless it buys the ability to address the new-characteristics
problem without any of the difficult econometric work involved in
solving estimated hedonic demand equations for reservation prices.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
Summary and Conclusion
Practical Pros and Cons
This is, we would argue, a practical plus.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
Summary and Conclusion
Practical Pros and Cons
This is, we would argue, a practical plus.
This is because NSI’s are never keen to adopt regression-based
methods when producing their inflation data because of (entirely valid)
worries over robustness and reproducibility.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
Summary and Conclusion
Practical Pros and Cons
This is, we would argue, a practical plus.
This is because NSI’s are never keen to adopt regression-based
methods when producing their inflation data because of (entirely valid)
worries over robustness and reproducibility.
Extrapolation of estimated hedonic models/demand equations might
therefore be a step too far.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
Summary and Conclusion
Practical Pros and Cons
This is, we would argue, a practical plus.
This is because NSI’s are never keen to adopt regression-based
methods when producing their inflation data because of (entirely valid)
worries over robustness and reproducibility.
Extrapolation of estimated hedonic models/demand equations might
therefore be a step too far.
Our method has the major advantage of not requiring NSI’s to do any
econometric work over-and-above what they are doing already.
Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41

More Related Content

Similar to Hedonic Models & New Characteristics

HOP-Rec_RecSys18
HOP-Rec_RecSys18HOP-Rec_RecSys18
HOP-Rec_RecSys18
Matt Yang
 
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
The Statistical and Applied Mathematical Sciences Institute
 
The importance of the Mix Methodology for Studying Offshore Energy sector
The importance of the Mix Methodology for Studying Offshore Energy sector The importance of the Mix Methodology for Studying Offshore Energy sector
The importance of the Mix Methodology for Studying Offshore Energy sector
inventionjournals
 
The Ascendance of the Dual Simplex Method: A Geometric View
The Ascendance of the Dual Simplex Method: A Geometric ViewThe Ascendance of the Dual Simplex Method: A Geometric View
The Ascendance of the Dual Simplex Method: A Geometric View
Bob Fourer
 
Business Economics - Unit-3 IMBA Syllabus Osmania University
Business Economics - Unit-3 IMBA Syllabus Osmania UniversityBusiness Economics - Unit-3 IMBA Syllabus Osmania University
Business Economics - Unit-3 IMBA Syllabus Osmania University
Balasri Kamarapu
 
Financial Benchmarking Of Transportation Companies In The New York Stock Exc...
Financial Benchmarking Of Transportation Companies  In The New York Stock Exc...Financial Benchmarking Of Transportation Companies  In The New York Stock Exc...
Financial Benchmarking Of Transportation Companies In The New York Stock Exc...
ertekg
 
FMI output gap
FMI output gapFMI output gap
FMI output gap
ManfredNolte
 
Wp13105
Wp13105Wp13105
Wp13105
ManfredNolte
 
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
ijccsa
 
Production and the Market Process, Lecture 4 with Peter Klein - Mises Academy
Production and the Market Process, Lecture 4 with Peter Klein - Mises AcademyProduction and the Market Process, Lecture 4 with Peter Klein - Mises Academy
Production and the Market Process, Lecture 4 with Peter Klein - Mises Academy
The Ludwig von Mises Institute
 
Reusing and integrating public proteomics data to improve our knowledge of th...
Reusing and integrating public proteomics data to improve our knowledge of th...Reusing and integrating public proteomics data to improve our knowledge of th...
Reusing and integrating public proteomics data to improve our knowledge of th...
Juan Antonio Vizcaino
 
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
pkconference
 
Pieroni Ricciarelli 2009
Pieroni Ricciarelli 2009Pieroni Ricciarelli 2009
Pieroni Ricciarelli 2009
ricciarellim
 
Growth curves by rance and sueno
Growth curves by rance and suenoGrowth curves by rance and sueno
Growth curves by rance and sueno
Mario Rance
 
EMPIRICAL PROJECTObjective to help students put in practice w.docx
EMPIRICAL PROJECTObjective  to help students put in practice w.docxEMPIRICAL PROJECTObjective  to help students put in practice w.docx
EMPIRICAL PROJECTObjective to help students put in practice w.docx
SALU18
 
Stochastic optimization and risk management for an efficient planning of buil...
Stochastic optimization and risk management for an efficient planning of buil...Stochastic optimization and risk management for an efficient planning of buil...
Stochastic optimization and risk management for an efficient planning of buil...
Emilio L. Cano
 
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
Alexander Decker
 

Similar to Hedonic Models & New Characteristics (17)

HOP-Rec_RecSys18
HOP-Rec_RecSys18HOP-Rec_RecSys18
HOP-Rec_RecSys18
 
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
CLIM Program: Remote Sensing Workshop, Statistical Emulation with Dimension R...
 
The importance of the Mix Methodology for Studying Offshore Energy sector
The importance of the Mix Methodology for Studying Offshore Energy sector The importance of the Mix Methodology for Studying Offshore Energy sector
The importance of the Mix Methodology for Studying Offshore Energy sector
 
The Ascendance of the Dual Simplex Method: A Geometric View
The Ascendance of the Dual Simplex Method: A Geometric ViewThe Ascendance of the Dual Simplex Method: A Geometric View
The Ascendance of the Dual Simplex Method: A Geometric View
 
Business Economics - Unit-3 IMBA Syllabus Osmania University
Business Economics - Unit-3 IMBA Syllabus Osmania UniversityBusiness Economics - Unit-3 IMBA Syllabus Osmania University
Business Economics - Unit-3 IMBA Syllabus Osmania University
 
Financial Benchmarking Of Transportation Companies In The New York Stock Exc...
Financial Benchmarking Of Transportation Companies  In The New York Stock Exc...Financial Benchmarking Of Transportation Companies  In The New York Stock Exc...
Financial Benchmarking Of Transportation Companies In The New York Stock Exc...
 
FMI output gap
FMI output gapFMI output gap
FMI output gap
 
Wp13105
Wp13105Wp13105
Wp13105
 
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
Towards An Enhanced Semantic Approach Based On Formal Concept Analysis And Li...
 
Production and the Market Process, Lecture 4 with Peter Klein - Mises Academy
Production and the Market Process, Lecture 4 with Peter Klein - Mises AcademyProduction and the Market Process, Lecture 4 with Peter Klein - Mises Academy
Production and the Market Process, Lecture 4 with Peter Klein - Mises Academy
 
Reusing and integrating public proteomics data to improve our knowledge of th...
Reusing and integrating public proteomics data to improve our knowledge of th...Reusing and integrating public proteomics data to improve our knowledge of th...
Reusing and integrating public proteomics data to improve our knowledge of th...
 
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
ntermediate Production in a Stock-Flow Consistent Model with Environmental Ex...
 
Pieroni Ricciarelli 2009
Pieroni Ricciarelli 2009Pieroni Ricciarelli 2009
Pieroni Ricciarelli 2009
 
Growth curves by rance and sueno
Growth curves by rance and suenoGrowth curves by rance and sueno
Growth curves by rance and sueno
 
EMPIRICAL PROJECTObjective to help students put in practice w.docx
EMPIRICAL PROJECTObjective  to help students put in practice w.docxEMPIRICAL PROJECTObjective  to help students put in practice w.docx
EMPIRICAL PROJECTObjective to help students put in practice w.docx
 
Stochastic optimization and risk management for an efficient planning of buil...
Stochastic optimization and risk management for an efficient planning of buil...Stochastic optimization and risk management for an efficient planning of buil...
Stochastic optimization and risk management for an efficient planning of buil...
 
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
Measuring the volatility in ghana’s gross domestic product (gdp) rate using t...
 

More from Office for National Statistics

Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptxSlideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
Office for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 22 April 2024.
SlideShare ONS Economic Forum Slidepack - 22 April 2024.SlideShare ONS Economic Forum Slidepack - 22 April 2024.
SlideShare ONS Economic Forum Slidepack - 22 April 2024.
Office for National Statistics
 
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptxSlideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
Office for National Statistics
 
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptxSlideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
Office for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 22 January 2024
SlideShare ONS Economic Forum Slidepack - 22 January 2024SlideShare ONS Economic Forum Slidepack - 22 January 2024
SlideShare ONS Economic Forum Slidepack - 22 January 2024
Office for National Statistics
 
Beyond GDP: international developments and emerging frameworks - 26 September...
Beyond GDP: international developments and emerging frameworks - 26 September...Beyond GDP: international developments and emerging frameworks - 26 September...
Beyond GDP: international developments and emerging frameworks - 26 September...
Office for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 11 December 2023
SlideShare ONS Economic Forum Slidepack - 11 December 2023SlideShare ONS Economic Forum Slidepack - 11 December 2023
SlideShare ONS Economic Forum Slidepack - 11 December 2023
Office for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 13 November 2023
SlideShare ONS Economic Forum Slidepack - 13 November 2023SlideShare ONS Economic Forum Slidepack - 13 November 2023
SlideShare ONS Economic Forum Slidepack - 13 November 2023
Office for National Statistics
 
SlideShare ONS Economic Forum Slidepack - 16 October 2023
SlideShare ONS Economic Forum Slidepack - 16 October 2023SlideShare ONS Economic Forum Slidepack - 16 October 2023
SlideShare ONS Economic Forum Slidepack - 16 October 2023
Office for National Statistics
 
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
Office for National Statistics
 
GDP after 2025: updating national accounts and balance of payments – 11 Octob...
GDP after 2025: updating national accounts and balance of payments – 11 Octob...GDP after 2025: updating national accounts and balance of payments – 11 Octob...
GDP after 2025: updating national accounts and balance of payments – 11 Octob...
Office for National Statistics
 
SlideShare Measuring the Economy Slidepack - 29 September 2023
SlideShare Measuring the Economy Slidepack - 29 September 2023SlideShare Measuring the Economy Slidepack - 29 September 2023
SlideShare Measuring the Economy Slidepack - 29 September 2023
Office for National Statistics
 
Why dashboards?
Why dashboards?Why dashboards?
SlideShare ONS Economic Forum Slidepack - 18 September 2023
SlideShare ONS Economic Forum Slidepack - 18 September 2023SlideShare ONS Economic Forum Slidepack - 18 September 2023
SlideShare ONS Economic Forum Slidepack - 18 September 2023
Office for National Statistics
 
Connecting to the StatXplore API in PowerBI
Connecting to the StatXplore API in PowerBIConnecting to the StatXplore API in PowerBI
Connecting to the StatXplore API in PowerBI
Office for National Statistics
 
ONS Local presents Suffolk County Council's Cost of Living Dashboard
ONS Local presents Suffolk County Council's Cost of Living DashboardONS Local presents Suffolk County Council's Cost of Living Dashboard
ONS Local presents Suffolk County Council's Cost of Living Dashboard
Office for National Statistics
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
Office for National Statistics
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
Office for National Statistics
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
Office for National Statistics
 
ONS Local presents: Adult Education Outcomes in London
ONS Local presents: Adult Education Outcomes in LondonONS Local presents: Adult Education Outcomes in London
ONS Local presents: Adult Education Outcomes in London
Office for National Statistics
 

More from Office for National Statistics (20)

Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptxSlideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 13 May 2024.pptx
 
SlideShare ONS Economic Forum Slidepack - 22 April 2024.
SlideShare ONS Economic Forum Slidepack - 22 April 2024.SlideShare ONS Economic Forum Slidepack - 22 April 2024.
SlideShare ONS Economic Forum Slidepack - 22 April 2024.
 
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptxSlideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 18 March 2024.pptx
 
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptxSlideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
Slideshare - ONS Economic Forum Slidepack - 19 February 2024.pptx
 
SlideShare ONS Economic Forum Slidepack - 22 January 2024
SlideShare ONS Economic Forum Slidepack - 22 January 2024SlideShare ONS Economic Forum Slidepack - 22 January 2024
SlideShare ONS Economic Forum Slidepack - 22 January 2024
 
Beyond GDP: international developments and emerging frameworks - 26 September...
Beyond GDP: international developments and emerging frameworks - 26 September...Beyond GDP: international developments and emerging frameworks - 26 September...
Beyond GDP: international developments and emerging frameworks - 26 September...
 
SlideShare ONS Economic Forum Slidepack - 11 December 2023
SlideShare ONS Economic Forum Slidepack - 11 December 2023SlideShare ONS Economic Forum Slidepack - 11 December 2023
SlideShare ONS Economic Forum Slidepack - 11 December 2023
 
SlideShare ONS Economic Forum Slidepack - 13 November 2023
SlideShare ONS Economic Forum Slidepack - 13 November 2023SlideShare ONS Economic Forum Slidepack - 13 November 2023
SlideShare ONS Economic Forum Slidepack - 13 November 2023
 
SlideShare ONS Economic Forum Slidepack - 16 October 2023
SlideShare ONS Economic Forum Slidepack - 16 October 2023SlideShare ONS Economic Forum Slidepack - 16 October 2023
SlideShare ONS Economic Forum Slidepack - 16 October 2023
 
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
So what does ‘Beyond GDP’ mean for the UK – 12 October 2023
 
GDP after 2025: updating national accounts and balance of payments – 11 Octob...
GDP after 2025: updating national accounts and balance of payments – 11 Octob...GDP after 2025: updating national accounts and balance of payments – 11 Octob...
GDP after 2025: updating national accounts and balance of payments – 11 Octob...
 
SlideShare Measuring the Economy Slidepack - 29 September 2023
SlideShare Measuring the Economy Slidepack - 29 September 2023SlideShare Measuring the Economy Slidepack - 29 September 2023
SlideShare Measuring the Economy Slidepack - 29 September 2023
 
Why dashboards?
Why dashboards?Why dashboards?
Why dashboards?
 
SlideShare ONS Economic Forum Slidepack - 18 September 2023
SlideShare ONS Economic Forum Slidepack - 18 September 2023SlideShare ONS Economic Forum Slidepack - 18 September 2023
SlideShare ONS Economic Forum Slidepack - 18 September 2023
 
Connecting to the StatXplore API in PowerBI
Connecting to the StatXplore API in PowerBIConnecting to the StatXplore API in PowerBI
Connecting to the StatXplore API in PowerBI
 
ONS Local presents Suffolk County Council's Cost of Living Dashboard
ONS Local presents Suffolk County Council's Cost of Living DashboardONS Local presents Suffolk County Council's Cost of Living Dashboard
ONS Local presents Suffolk County Council's Cost of Living Dashboard
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
 
ONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIsONS Local and Data Science Community Workshop 1: How to use APIs
ONS Local and Data Science Community Workshop 1: How to use APIs
 
ONS Local presents: Adult Education Outcomes in London
ONS Local presents: Adult Education Outcomes in LondonONS Local presents: Adult Education Outcomes in London
ONS Local presents: Adult Education Outcomes in London
 

Recently uploaded

一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
74nqk8xf
 
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
v7oacc3l
 
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
Social Samosa
 
Learn SQL from basic queries to Advance queries
Learn SQL from basic queries to Advance queriesLearn SQL from basic queries to Advance queries
Learn SQL from basic queries to Advance queries
manishkhaire30
 
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Aggregage
 
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
g4dpvqap0
 
End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024
Lars Albertsson
 
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
74nqk8xf
 
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
g4dpvqap0
 
DSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelinesDSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelines
Timothy Spann
 
Influence of Marketing Strategy and Market Competition on Business Plan
Influence of Marketing Strategy and Market Competition on Business PlanInfluence of Marketing Strategy and Market Competition on Business Plan
Influence of Marketing Strategy and Market Competition on Business Plan
jerlynmaetalle
 
University of New South Wales degree offer diploma Transcript
University of New South Wales degree offer diploma TranscriptUniversity of New South Wales degree offer diploma Transcript
University of New South Wales degree offer diploma Transcript
soxrziqu
 
The Ipsos - AI - Monitor 2024 Report.pdf
The  Ipsos - AI - Monitor 2024 Report.pdfThe  Ipsos - AI - Monitor 2024 Report.pdf
The Ipsos - AI - Monitor 2024 Report.pdf
Social Samosa
 
Intelligence supported media monitoring in veterinary medicine
Intelligence supported media monitoring in veterinary medicineIntelligence supported media monitoring in veterinary medicine
Intelligence supported media monitoring in veterinary medicine
AndrzejJarynowski
 
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
nyfuhyz
 
A presentation that explain the Power BI Licensing
A presentation that explain the Power BI LicensingA presentation that explain the Power BI Licensing
A presentation that explain the Power BI Licensing
AlessioFois2
 
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
nuttdpt
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
sameer shah
 
一比一原版(UO毕业证)渥太华大学毕业证如何办理
一比一原版(UO毕业证)渥太华大学毕业证如何办理一比一原版(UO毕业证)渥太华大学毕业证如何办理
一比一原版(UO毕业证)渥太华大学毕业证如何办理
aqzctr7x
 
Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...
Bill641377
 

Recently uploaded (20)

一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
一比一原版(Coventry毕业证书)考文垂大学毕业证如何办理
 
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
 
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
 
Learn SQL from basic queries to Advance queries
Learn SQL from basic queries to Advance queriesLearn SQL from basic queries to Advance queries
Learn SQL from basic queries to Advance queries
 
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
 
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
 
End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024
 
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
一比一原版(Chester毕业证书)切斯特大学毕业证如何办理
 
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
一比一原版(Glasgow毕业证书)格拉斯哥大学毕业证如何办理
 
DSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelinesDSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelines
 
Influence of Marketing Strategy and Market Competition on Business Plan
Influence of Marketing Strategy and Market Competition on Business PlanInfluence of Marketing Strategy and Market Competition on Business Plan
Influence of Marketing Strategy and Market Competition on Business Plan
 
University of New South Wales degree offer diploma Transcript
University of New South Wales degree offer diploma TranscriptUniversity of New South Wales degree offer diploma Transcript
University of New South Wales degree offer diploma Transcript
 
The Ipsos - AI - Monitor 2024 Report.pdf
The  Ipsos - AI - Monitor 2024 Report.pdfThe  Ipsos - AI - Monitor 2024 Report.pdf
The Ipsos - AI - Monitor 2024 Report.pdf
 
Intelligence supported media monitoring in veterinary medicine
Intelligence supported media monitoring in veterinary medicineIntelligence supported media monitoring in veterinary medicine
Intelligence supported media monitoring in veterinary medicine
 
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
一比一原版(UMN文凭证书)明尼苏达大学毕业证如何办理
 
A presentation that explain the Power BI Licensing
A presentation that explain the Power BI LicensingA presentation that explain the Power BI Licensing
A presentation that explain the Power BI Licensing
 
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
 
一比一原版(UO毕业证)渥太华大学毕业证如何办理
一比一原版(UO毕业证)渥太华大学毕业证如何办理一比一原版(UO毕业证)渥太华大学毕业证如何办理
一比一原版(UO毕业证)渥太华大学毕业证如何办理
 
Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...
 

Hedonic Models & New Characteristics

  • 1. Hedonic Models and New Characteristics Ian Crawford Oxford and IFS J. Peter Neary Oxford, CEPR and CESifo Challenges and Opportunities in Price Measurement The Shard July 19, 2018 Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 1 / 41
  • 2. Introduction Outline How to correct measured inflation for quality change? Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 3. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 4. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. But the standard method does not account for changes to product specifications on the extensive margin. This can lead to bias. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 5. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. But the standard method does not account for changes to product specifications on the extensive margin. This can lead to bias. We show how to correct for it and provide an empirical example. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 6. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. But the standard method does not account for changes to product specifications on the extensive margin. This can lead to bias. We show how to correct for it and provide an empirical example. Our methodological innovation is to combine: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 7. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. But the standard method does not account for changes to product specifications on the extensive margin. This can lead to bias. We show how to correct for it and provide an empirical example. Our methodological innovation is to combine: Linear characteristics model of Gorman (1956) Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 8. Introduction Outline How to correct measured inflation for quality change? National statistical agencies, like the ONS or the BLS or Eurostat, now routinely use “hedonic” methods to correct their measures of inflation. But the standard method does not account for changes to product specifications on the extensive margin. This can lead to bias. We show how to correct for it and provide an empirical example. Our methodological innovation is to combine: Linear characteristics model of Gorman (1956) Sato (1976) - Vartia (1976) - Feenstra (1994) index number for new goods Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 2 / 41
  • 9. Introduction Outline 1 Background Theory 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 3 / 41
  • 10. Background Theory Outline 1 Background Theory Measuring Inflation Hedonic Methods The Linear Characteristics Model 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 4 / 41
  • 11. Background Theory Measuring Inflation The True Cost-of-Living Index A utility-maximizing consumer: x1 t , ..., xK t = arg max x1,...,xK u(x1 , ..., xK ) ∑ k pk t xk = yt Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
  • 12. Background Theory Measuring Inflation The True Cost-of-Living Index A utility-maximizing consumer: x1 t , ..., xK t = arg max x1,...,xK u(x1 , ..., xK ) ∑ k pk t xk = yt The expenditure function: c(pt, u) = min x1,...,xK ∑ j p j txj u(x1 , ..., xK ) = u Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
  • 13. Background Theory Measuring Inflation The True Cost-of-Living Index A utility-maximizing consumer: x1 t , ..., xK t = arg max x1,...,xK u(x1 , ..., xK ) ∑ k pk t xk = yt The expenditure function: c(pt, u) = min x1,...,xK ∑ j p j txj u(x1 , ..., xK ) = u The constant-utility Kon¨us price index is defined as: PK(p0, p1, u) = c(p1, u) c(p0, u) Characteristics New Characteristics Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 5 / 41
  • 14. Background Theory Measuring Inflation Observable Price Indexes Standard index number theory then provides a number of ways of either approximating or, subject to functional form assumptions, exactly computing the true index. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
  • 15. Background Theory Measuring Inflation Observable Price Indexes Standard index number theory then provides a number of ways of either approximating or, subject to functional form assumptions, exactly computing the true index. For example: PL = ∑ j p j 1 p j 0 s j 0 PP =  ∑ j p j 1 p j 0 −1 s j 1   −1 PF = PLPP PT = ∏ j p j 1 p j 0 1 2 (s j 0+s j 1) where s j t = p j tx j t ∑i pi txi t . Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
  • 16. Background Theory Measuring Inflation Observable Price Indexes Standard index number theory then provides a number of ways of either approximating or, subject to functional form assumptions, exactly computing the true index. For example: PL = ∑ j p j 1 p j 0 s j 0 PP =  ∑ j p j 1 p j 0 −1 s j 1   −1 PF = PLPP PT = ∏ j p j 1 p j 0 1 2 (s j 0+s j 1) where s j t = p j tx j t ∑i pi txi t . Problem: No allowance for quality change Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 6 / 41
  • 17. Background Theory Hedonic Methods Hedonic Methods There are two basic methods of correcting for quality change: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
  • 18. Background Theory Hedonic Methods Hedonic Methods There are two basic methods of correcting for quality change: 1 the time dummy method - based on the repackaging model of Fisher and Shell (1971) and Muellbauer (1974) Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
  • 19. Background Theory Hedonic Methods Hedonic Methods There are two basic methods of correcting for quality change: 1 the time dummy method - based on the repackaging model of Fisher and Shell (1971) and Muellbauer (1974) 2 the indirect method - based on the linear characteristics model of Gorman (1956) and Lancaster (1966). Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 7 / 41
  • 20. Background Theory Hedonic Methods The Linear Characteristics Model There are K varieties of products with quantities denoted by xk and unit prices by pk. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
  • 21. Background Theory Hedonic Methods The Linear Characteristics Model There are K varieties of products with quantities denoted by xk and unit prices by pk. These products are differentiated by J characteristics. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
  • 22. Background Theory Hedonic Methods The Linear Characteristics Model There are K varieties of products with quantities denoted by xk and unit prices by pk. These products are differentiated by J characteristics. In the linear characteristics model the total amount of a given characteristic present in a bundle of varieties x1 t , ..., xK t observed in period t is z j t = ∑ k a k,j t xk t where a k,j t represents the amount of characteristic j present in one unit of product k according to the product specification in period t. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 8 / 41
  • 23. Background Theory The Linear Characteristics Model The Linear Characteristics Model The consumer’s problem is: max x1,...,xK v(z1 , ..., zJ ) subject to: ∑ k pk t xk = yt and zj = ∑ k a k,j t xk Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
  • 24. Background Theory The Linear Characteristics Model The Linear Characteristics Model The consumer’s problem is: max x1,...,xK v(z1 , ..., zJ ) subject to: ∑ k pk t xk = yt and zj = ∑ k a k,j t xk Maximising behaviour implies the first order conditions: pk t = ∑ j a k,j t π j t ∀k, t where π j t ≡ 1 ψt v j t are the shadow prices of characteristics. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
  • 25. Background Theory The Linear Characteristics Model The Linear Characteristics Model The consumer’s problem is: max x1,...,xK v(z1 , ..., zJ ) subject to: ∑ k pk t xk = yt and zj = ∑ k a k,j t xk Maximising behaviour implies the first order conditions: pk t = ∑ j a k,j t π j t ∀k, t where π j t ≡ 1 ψt v j t are the shadow prices of characteristics. This is the hedonic pricing equation. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 9 / 41
  • 26. Background Theory The Linear Characteristics Model The Linear Characteristics Model The budget constraint “adds up” in both characteristics and product space: ∑ k pk t xk t = ∑ j π j tz j t = yt Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 10 / 41
  • 27. Background Theory The Linear Characteristics Model The Linear Characteristics Model The budget constraint “adds up” in both characteristics and product space: ∑ k pk t xk t = ∑ j π j tz j t = yt The model can be thought of in terms of either preferences- for-products or preferences-for-characteristics: x1 t , ..., xK t = arg max u(x1 , ..., xK ) ∑ k pk t xk = yt z1 t , ..., zJ t = arg max v(z1 , ..., zJ ) ∑ j π j tzj = yt Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 10 / 41
  • 28. Background Theory The Linear Characteristics Model The Linear Characteristics Model z1 t , ..., zJ t = arg max v(z1 , ..., zJ ) ∑ j π j tzj = yt The hedonic expenditure function: c(πt, v) = min z1,...,zJ ∑ j π j tz j t v(z1 , ..., zJ ) = v Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 11 / 41
  • 29. Background Theory The Linear Characteristics Model The Linear Characteristics Model z1 t , ..., zJ t = arg max v(z1 , ..., zJ ) ∑ j π j tzj = yt The hedonic expenditure function: c(πt, v) = min z1,...,zJ ∑ j π j tz j t v(z1 , ..., zJ ) = v The constant-utility Kon¨us hedonic price index is defined as: PK(π0, π1, v) = c(π1, v) c(π0, v) CostOfLiving New Characteristics Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 11 / 41
  • 30. Background Theory The Linear Characteristics Model The Linear Characteristics Model All of these hedonic price indices depend on shadow price-relatives: π j 1 π j 0 Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
  • 31. Background Theory The Linear Characteristics Model The Linear Characteristics Model All of these hedonic price indices depend on shadow price-relatives: π j 1 π j 0 If a new feature is added to products in this category in period 1: what is the denominator π j 0? Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
  • 32. Background Theory The Linear Characteristics Model The Linear Characteristics Model All of these hedonic price indices depend on shadow price-relatives: π j 1 π j 0 If a new feature is added to products in this category in period 1: what is the denominator π j 0? If an existing feature is withdrawn from the products in this category in period 1: what is the numerator π j 1? Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 12 / 41
  • 33. Adjusting for New Characteristics: General Outline 1 Background Theory 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 13 / 41
  • 34. Adjusting for New Characteristics: General Rationing Theory The solution comes from the theory of consumer behaviour under rationing (Hicks (1940), Rothbarth (1941), Neary and Roberts (1980)) Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
  • 35. Adjusting for New Characteristics: General Rationing Theory The solution comes from the theory of consumer behaviour under rationing (Hicks (1940), Rothbarth (1941), Neary and Roberts (1980)) Central idea: Use “virtual prices” ˜π Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
  • 36. Adjusting for New Characteristics: General Rationing Theory The solution comes from the theory of consumer behaviour under rationing (Hicks (1940), Rothbarth (1941), Neary and Roberts (1980)) Central idea: Use “virtual prices” ˜π Prices that would induce the consumer to consume voluntarily the constrained characteristic Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
  • 37. Adjusting for New Characteristics: General Rationing Theory The solution comes from the theory of consumer behaviour under rationing (Hicks (1940), Rothbarth (1941), Neary and Roberts (1980)) Central idea: Use “virtual prices” ˜π Prices that would induce the consumer to consume voluntarily the constrained characteristic Here, the “constraint” is that a characteristic is unavailable in one period Skip Details Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 14 / 41
  • 38. Adjusting for New Characteristics: General Rationing Theory Consider the constrained hedonic cost function ˜c(πt, v, ¯z1 t ) = min z2,...,zJ π1 t ¯z1 t + ∑ j=2 π j tzj subject to v(¯z1 , ..., zJ ) = v Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 15 / 41
  • 39. Adjusting for New Characteristics: General Rationing Theory Consider the constrained hedonic cost function ˜c(πt, v, ¯z1 t ) = min z2,...,zJ π1 t ¯z1 t + ∑ j=2 π j tzj subject to v(¯z1 , ..., zJ ) = v It satisfies Sheppard’s Lemma π1 ˜c(πt, v, ¯z1 t ) = ¯z1 t πj ˜c(πt, v, ¯z1 t ) = zj (πt, v, ¯z1 t ) ∀j = 2, ..J Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 15 / 41
  • 40. Adjusting for New Characteristics: General New Characteristics Bias The relationship between the constrained and unconstrained hedonic cost functions is given by ˜c(πt, v, ¯z1 t ) = π1 t ¯z1 t + ∑ j=2 π j tzj (πt, v, ¯z1 t ) = π1 t z1 ( ˜πt, v) + ∑ j=2 π j tzj ( ˜πt, v) = c( ˜πt, v) + π1 t − ˜π1 t ¯z1 t where ˜π1 t is the virtual price of the constrained characteristic and ˜πt = ˜π1 t , π2 t , ..., πJ t is the shadow support price vector. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 16 / 41
  • 41. Adjusting for New Characteristics: General New Characteristics Bias The shadow support prices are those which would precisely induce the consumer to choose the constrained characteristics vector ¯z1 t = z1 ( ˜πt, v) Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
  • 42. Adjusting for New Characteristics: General New Characteristics Bias The shadow support prices are those which would precisely induce the consumer to choose the constrained characteristics vector ¯z1 t = z1 ( ˜πt, v) The welfare cost of the constraint on the characteristic is ∂˜c(πt, v, ¯z1 t ) ∂¯z1 t = ∂c( ˜πt, v) ∂π1 t ∂π1 t (πt, v, ¯z1 t ) ∂¯z1 t − ∂π1 t (πt, v, ¯z1 t ) ∂¯z1 t ¯z1 t + π1 t − ˜π1 t = π1 t − ˜π1 t Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
  • 43. Adjusting for New Characteristics: General New Characteristics Bias The shadow support prices are those which would precisely induce the consumer to choose the constrained characteristics vector ¯z1 t = z1 ( ˜πt, v) The welfare cost of the constraint on the characteristic is ∂˜c(πt, v, ¯z1 t ) ∂¯z1 t = ∂c( ˜πt, v) ∂π1 t ∂π1 t (πt, v, ¯z1 t ) ∂¯z1 t − ∂π1 t (πt, v, ¯z1 t ) ∂¯z1 t ¯z1 t + π1 t − ˜π1 t = π1 t − ˜π1 t Thus the change in welfare associated with a change in the constraint is the difference between the shadow price of the characteristic and its virtual counterpart. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 17 / 41
  • 44. Adjusting for New Characteristics: General New Characteristics Bias Consider two periods t ∈ {0, 1} and suppose that the first characteristic is not available in the base period ¯z1 0 = 0 but is added to the product specification in the following period. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
  • 45. Adjusting for New Characteristics: General New Characteristics Bias Consider two periods t ∈ {0, 1} and suppose that the first characteristic is not available in the base period ¯z1 0 = 0 but is added to the product specification in the following period. Then ˜c(π0, v, ¯z1 0) = c( ˜π0, v) + π1 0 − ˜π1 t ¯z1 0 which becomes ˜c(π0, v, 0) = c( ˜π0, v). Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
  • 46. Adjusting for New Characteristics: General New Characteristics Bias Consider two periods t ∈ {0, 1} and suppose that the first characteristic is not available in the base period ¯z1 0 = 0 but is added to the product specification in the following period. Then ˜c(π0, v, ¯z1 0) = c( ˜π0, v) + π1 0 − ˜π1 t ¯z1 0 which becomes ˜c(π0, v, 0) = c( ˜π0, v). Consequently, the Kon¨us hedonic cost-of-living index can be expressed conveniently as the ratio of unconstrained hedonic cost functions where the base prices are given by the virtual prices: PK( ˜π0, π1, v) = c(π1, v) c( ˜π0, v) CostOfLiving Characteristics Sato-Vartia Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 18 / 41
  • 47. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 48. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. But all of them require that the shadow price relative π1 1/ ˜π1 0 is known. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 49. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. But all of them require that the shadow price relative π1 1/ ˜π1 0 is known. Solutions: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 50. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. But all of them require that the shadow price relative π1 1/ ˜π1 0 is known. Solutions: Ignore the affected characteristics and chain. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 51. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. But all of them require that the shadow price relative π1 1/ ˜π1 0 is known. Solutions: Ignore the affected characteristics and chain. Estimate the model and solve 0 = z1(π, v) for ˜π1. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 52. Adjusting for New Characteristics: General New Characteristics Bias The fact that everything can be expressed in terms of the unconstrained cost function means that all of the standard index number results are available. But all of them require that the shadow price relative π1 1/ ˜π1 0 is known. Solutions: Ignore the affected characteristics and chain. Estimate the model and solve 0 = z1(π, v) for ˜π1. Something else ... Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 19 / 41
  • 53. Adjusting for New Characteristics: A Structural Approach Outline 1 Background Theory 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 20 / 41
  • 54. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index To make progress, we need to impose more structure Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
  • 55. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index To make progress, we need to impose more structure We assume CES preferences over characteristics: v(z) = ∑ βjz σ−1 σ j σ σ−1 = ∑ βjzθ j 1 θ c(π, u) = c(π)u = ∑ βσ j π1−σ j 1 1−σ u Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
  • 56. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index To make progress, we need to impose more structure We assume CES preferences over characteristics: v(z) = ∑ βjz σ−1 σ j σ σ−1 = ∑ βjzθ j 1 θ c(π, u) = c(π)u = ∑ βσ j π1−σ j 1 1−σ u where σ ≡ 1 1−θ and θ ≡ σ−1 σ . Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 21 / 41
  • 57. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index The Sato-Vartia cost-of-living index is exact for CES preferences Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
  • 58. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index The Sato-Vartia cost-of-living index is exact for CES preferences This holds for preferences over characteristics as well as over goods: PSV = c(π1) c(π0) = ∏ j πj,1 πj,0 ωj where: ωj = µj ∑ µj µj = sj,1 − sj,0 ln sj,1 − ln sj,0 Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
  • 59. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia Index The Sato-Vartia cost-of-living index is exact for CES preferences This holds for preferences over characteristics as well as over goods: PSV = c(π1) c(π0) = ∏ j πj,1 πj,0 ωj where: ωj = µj ∑ µj µj = sj,1 − sj,0 ln sj,1 − ln sj,0 Note that this corresponds exactly to the true hedonic index but does not depend on unobserved coefficients. New Characteristics Sato-Vartia-Feenstra Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 22 / 41
  • 60. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index Now consider two periods t ∈ {0, 1} and suppose that the set of characteristics changes from J0 in the base period to J1 in the comparison period. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
  • 61. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index Now consider two periods t ∈ {0, 1} and suppose that the set of characteristics changes from J0 in the base period to J1 in the comparison period. The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics which are available in both periods. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
  • 62. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index Now consider two periods t ∈ {0, 1} and suppose that the set of characteristics changes from J0 in the base period to J1 in the comparison period. The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics which are available in both periods. We can show that: PSVF = c( ˜π1) c( ˜π0) = λ1 λ0 1 σ−1 ∏ j∈J π j 1 π j 0 ωj is the exact price index. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
  • 63. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index Now consider two periods t ∈ {0, 1} and suppose that the set of characteristics changes from J0 in the base period to J1 in the comparison period. The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics which are available in both periods. We can show that: PSVF = c( ˜π1) c( ˜π0) = λ1 λ0 1 σ−1 ∏ j∈J π j 1 π j 0 ωj is the exact price index. It equals the Sato-Vartia index, times an adjustment factor Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
  • 64. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index Now consider two periods t ∈ {0, 1} and suppose that the set of characteristics changes from J0 in the base period to J1 in the comparison period. The intersection J = J0 ∩ J1 = ∅ contains the set of characteristics which are available in both periods. We can show that: PSVF = c( ˜π1) c( ˜π0) = λ1 λ0 1 σ−1 ∏ j∈J π j 1 π j 0 ωj is the exact price index. It equals the Sato-Vartia index, times an adjustment factor This needs an estimate of σ, but otherwise depends only on observables. Sato-Vartia Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 23 / 41
  • 65. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index The adjustment factor λ1 λ0 1 σ−1 where λt = ∑j∈J π j tz j t ∑j∈Jt π j tz j t ≤ 1 takes into account the changing set of characteristics on the extensive margin. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
  • 66. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index The adjustment factor λ1 λ0 1 σ−1 where λt = ∑j∈J π j tz j t ∑j∈Jt π j tz j t ≤ 1 takes into account the changing set of characteristics on the extensive margin. If new characteristics are important, λ1 will tend to be small so the hedonic price index will be lower. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
  • 67. Adjusting for New Characteristics: A Structural Approach The Sato-Vartia-Feenstra Index The adjustment factor λ1 λ0 1 σ−1 where λt = ∑j∈J π j tz j t ∑j∈Jt π j tz j t ≤ 1 takes into account the changing set of characteristics on the extensive margin. If new characteristics are important, λ1 will tend to be small so the hedonic price index will be lower. The adjustment factor is less important the higher is the substitution parameter σ, i.e., ignoring new characteristics matters less if they are close substitutes for existing ones. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 24 / 41
  • 68. An Application: New Cars, New Characteristics Outline 1 Background Theory 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 25 / 41
  • 69. An Application: New Cars, New Characteristics The Data Data were collected by the UK Office for National Statistics from What Car on the major new car models available from 1988 to mid 1995. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
  • 70. An Application: New Cars, New Characteristics The Data Data were collected by the UK Office for National Statistics from What Car on the major new car models available from 1988 to mid 1995. Data on market shares and sales are from the Competition Commission. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
  • 71. An Application: New Cars, New Characteristics The Data Data were collected by the UK Office for National Statistics from What Car on the major new car models available from 1988 to mid 1995. Data on market shares and sales are from the Competition Commission. The data covers 36 main models. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
  • 72. An Application: New Cars, New Characteristics The Data Data were collected by the UK Office for National Statistics from What Car on the major new car models available from 1988 to mid 1995. Data on market shares and sales are from the Competition Commission. The data covers 36 main models. Prices and specifications are measured quarterly (30 periods, 3160 observations) Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 26 / 41
  • 73. An Application: New Cars, New Characteristics Summary Statistics Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 27 / 41
  • 74. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 75. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 76. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; and at the beginning of the second quarter, which normally coincides with changes in vehicle excise duty and taxes on motor fuels. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 77. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; and at the beginning of the second quarter, which normally coincides with changes in vehicle excise duty and taxes on motor fuels. Looking at the intensive margin first, there were quality improvements in terms of performance across the board: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 78. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; and at the beginning of the second quarter, which normally coincides with changes in vehicle excise duty and taxes on motor fuels. Looking at the intensive margin first, there were quality improvements in terms of performance across the board: 0-60 mph acceleration times improved over the period by 6.5% Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 79. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; and at the beginning of the second quarter, which normally coincides with changes in vehicle excise duty and taxes on motor fuels. Looking at the intensive margin first, there were quality improvements in terms of performance across the board: 0-60 mph acceleration times improved over the period by 6.5% fuel economy improved by 2.8%. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 80. An Application: New Cars, New Characteristics Changes in Characteristics Specifications generally change twice per year: at the beginning of the fourth quarter, which marks the time when new annual registration letters were issued in the UK; and at the beginning of the second quarter, which normally coincides with changes in vehicle excise duty and taxes on motor fuels. Looking at the intensive margin first, there were quality improvements in terms of performance across the board: 0-60 mph acceleration times improved over the period by 6.5% fuel economy improved by 2.8%. both torque and brake-horse-power improved by 5.2% on average (torque and bhp are proportional at a fixed engine speed). Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 28 / 41
  • 81. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 82. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: central locking was present in a quarter of cars at the start of the period; this rose to nearly two-thirds by the end, Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 83. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: central locking was present in a quarter of cars at the start of the period; this rose to nearly two-thirds by the end, manual sun-roofs went from being available on 17.6% of models to 25% Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 84. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: central locking was present in a quarter of cars at the start of the period; this rose to nearly two-thirds by the end, manual sun-roofs went from being available on 17.6% of models to 25% electric front-windows were uncommon at beginning with 8% market share, but by the end 40% of new models had them. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 85. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: central locking was present in a quarter of cars at the start of the period; this rose to nearly two-thirds by the end, manual sun-roofs went from being available on 17.6% of models to 25% electric front-windows were uncommon at beginning with 8% market share, but by the end 40% of new models had them. electrically operated door mirrors went from 8% to 25% Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 86. An Application: New Cars, New Characteristics Changes in Characteristics The availability of features also improved: central locking was present in a quarter of cars at the start of the period; this rose to nearly two-thirds by the end, manual sun-roofs went from being available on 17.6% of models to 25% electric front-windows were uncommon at beginning with 8% market share, but by the end 40% of new models had them. electrically operated door mirrors went from 8% to 25% split-fold rear seats increased from being specified in around half of new cars to nearly three-quarters. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 29 / 41
  • 87. An Application: New Cars, New Characteristics Changes in Characteristics On the extensive margin there were a number of changes to the set of characteristics over the period. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
  • 88. An Application: New Cars, New Characteristics Changes in Characteristics On the extensive margin there were a number of changes to the set of characteristics over the period. electronic seat and steering-adjustability both became available in the fourth quarter of 1988; Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
  • 89. An Application: New Cars, New Characteristics Changes in Characteristics On the extensive margin there were a number of changes to the set of characteristics over the period. electronic seat and steering-adjustability both became available in the fourth quarter of 1988; electric sun-roofs were introduced in the first quarter of 1990; Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
  • 90. An Application: New Cars, New Characteristics Changes in Characteristics On the extensive margin there were a number of changes to the set of characteristics over the period. electronic seat and steering-adjustability both became available in the fourth quarter of 1988; electric sun-roofs were introduced in the first quarter of 1990; air conditioning was introduced in the fourth quarter of 1993 as were powered (heated) seats and drivers’ airbags. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
  • 91. An Application: New Cars, New Characteristics Changes in Characteristics On the extensive margin there were a number of changes to the set of characteristics over the period. electronic seat and steering-adjustability both became available in the fourth quarter of 1988; electric sun-roofs were introduced in the first quarter of 1990; air conditioning was introduced in the fourth quarter of 1993 as were powered (heated) seats and drivers’ airbags. CD-players were introduced in the second quarter of 1994. By then simple radio-only models had gone - they disappeared from the new car market by the fourth quarter of 1992. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 30 / 41
  • 92. An Application: New Cars, New Characteristics Changes in Characteristics Other features were short-lived and simply came and went: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
  • 93. An Application: New Cars, New Characteristics Changes in Characteristics Other features were short-lived and simply came and went: head-lamp cleaners were available in some cars (Volvos) at the start of our period of observation, but disappeared at the start of 1990 only to reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and Volvo’s again) before disappearing again. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
  • 94. An Application: New Cars, New Characteristics Changes in Characteristics Other features were short-lived and simply came and went: head-lamp cleaners were available in some cars (Volvos) at the start of our period of observation, but disappeared at the start of 1990 only to reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and Volvo’s again) before disappearing again. Similarly early trip-computers became available (also on Alfa’s) but disappeared only to be offered briefly by Rover before again disappearing. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
  • 95. An Application: New Cars, New Characteristics Changes in Characteristics Other features were short-lived and simply came and went: head-lamp cleaners were available in some cars (Volvos) at the start of our period of observation, but disappeared at the start of 1990 only to reappear briefly from mid 1993 to mid 1994 (on Alfa Romeo’s and Volvo’s again) before disappearing again. Similarly early trip-computers became available (also on Alfa’s) but disappeared only to be offered briefly by Rover before again disappearing. Trip computers are now completely standard but they didn’t take off until after the end of our period of observation. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 31 / 41
  • 96. An Application: New Cars, New Characteristics The Profile of New Car Prices The density of the prices of new cars over time 5 10 15 20 0.000.050.100.150.20 £ '000's Density Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 32 / 41
  • 97. An Application: New Cars, New Characteristics The Profile of New Car Prices The density of the prices of new cars over time 88 90 92 94 05101520 £'000's 88 89 90 91 92 93 94 95 200 200 200 200 200 400 400 400 400 400 400 600 600 600 600 600 800 800 1000 1000 1000 1200 1200 12001400 1400 1600 1600 1600 1600 1800 1800 1800 2000 2000 2000 Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 33 / 41
  • 98. An Application: New Cars, New Characteristics Recap on Methodology 1 Estimate within period linear characteristics models for each period using only extant characteristics to recover {πt, zt}t=0,...,T Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
  • 99. An Application: New Cars, New Characteristics Recap on Methodology 1 Estimate within period linear characteristics models for each period using only extant characteristics to recover {πt, zt}t=0,...,T 2 Fit a value for σ. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
  • 100. An Application: New Cars, New Characteristics Recap on Methodology 1 Estimate within period linear characteristics models for each period using only extant characteristics to recover {πt, zt}t=0,...,T 2 Fit a value for σ. 3 Compute the adjustment factors {λt}t=0,...,T Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
  • 101. An Application: New Cars, New Characteristics Recap on Methodology 1 Estimate within period linear characteristics models for each period using only extant characteristics to recover {πt, zt}t=0,...,T 2 Fit a value for σ. 3 Compute the adjustment factors {λt}t=0,...,T 4 Compute the variety-adjusted Sato-Vartia hedonic price index c(πt) c(π0) = λt λ0 1 σ−1 ∏ j∈J πj,t πj,0 ωj Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 34 / 41
  • 102. An Application: New Cars, New Characteristics Empirical Results: Price Levels 88 90 92 94 0.81.01.21.4 Index 88 89 90 91 92 93 94 95 Laspeyres Paasche Fisher Sato−Vartia Sato−Vartia−Feenstra Repackaging Notes: 1988q4: adjustable seats and adjustable steering introduced; 1990q1: electric sun roofs introduced; 1990q2: headlamp cleaners withdrawn; 1991q1: trip computers introduced; 1992q1: trip computers withdrawn; 1992q3: trip computers re-introduced for one quarter; 1992q4 radio-only withdrawn; 1993q2: headlamp cleaners re-introduced; 1993q4: air conditioning, power seats and driver’s airbags introduced; 1994q2 CD players available; 1993q4: headlamp cleaners finally withdrawn. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 35 / 41
  • 103. An Application: New Cars, New Characteristics Empirical Results: Inflation Rates 89 90 91 92 93 94 95 −0.3−0.2−0.10.00.1 Index 89 90 91 92 93 94 95 Laspeyres Paasche Fisher Sato−Vartia Sato−Vartia−Feenstra Repackaging Notes: 1988q4: adjustable seats and adjustable steering introduced; 1990q1: electric sun roofs introduced; 1990q2: headlamp cleaners withdrawn; 1991q1: trip computers introduced; 1992q1: trip computers withdrawn; 1992q3: trip computers re-introduced for one quarter; 1992q4 radio-only withdrawn; 1993q2: headlamp cleaners re-introduced; 1993q4: air conditioning, power seats and driver’s airbags introduced; 1994q2 CD players available; 1993q4: headlamp cleaners finally withdrawn. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 36 / 41
  • 104. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 105. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Hedonic quality adjustment on the basis of overlapping characteristics reduces this to an annual average growth rate of 3.45%. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 106. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Hedonic quality adjustment on the basis of overlapping characteristics reduces this to an annual average growth rate of 3.45%. Allowing for the changes on the extensive margin of characteristics gives an annual average growth rate of about 2%. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 107. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Hedonic quality adjustment on the basis of overlapping characteristics reduces this to an annual average growth rate of 3.45%. Allowing for the changes on the extensive margin of characteristics gives an annual average growth rate of about 2%. Very roughly: Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 108. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Hedonic quality adjustment on the basis of overlapping characteristics reduces this to an annual average growth rate of 3.45%. Allowing for the changes on the extensive margin of characteristics gives an annual average growth rate of about 2%. Very roughly: if we take the Boskin Commission’s estimate of the bias from quality change of 0.6 percentage points per year, Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 109. An Application: New Cars, New Characteristics Conclusions Without quality adjustment we see an annual average growth in prices of new cars of about 5.37%. Hedonic quality adjustment on the basis of overlapping characteristics reduces this to an annual average growth rate of 3.45%. Allowing for the changes on the extensive margin of characteristics gives an annual average growth rate of about 2%. Very roughly: if we take the Boskin Commission’s estimate of the bias from quality change of 0.6 percentage points per year, our results suggest that allowing for changes in the extensive margin could add half as much again to that. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 37 / 41
  • 110. Summary and Conclusion Outline 1 Background Theory 2 Adjusting for New Characteristics: General 3 Adjusting for New Characteristics: A Structural Approach 4 An Application: New Cars, New Characteristics 5 Summary and Conclusion Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 38 / 41
  • 111. Summary and Conclusion Summary We introduce a new approach to correcting inflation for quality change at the extensive margin The theory builds on and extends: The Gorman approach to preferences over characteristics The Feenstra approach to correcting price indices for new goods The method we propose is easy to apply in practice An application to UK new car prices shows that accounting for changes in characteristics matters: It reduces inflation by 1.45% per annum Half as much as accounting for quality change at the intensive margin Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 39 / 41
  • 112. Summary and Conclusion Practical Pros and Cons The main drawback of our approach is that it is based on an assumption. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
  • 113. Summary and Conclusion Practical Pros and Cons The main drawback of our approach is that it is based on an assumption. The generalised CES is “not-quite-superlative”, and therefore we cannot just assume that the approximation to true preferences will be good in the vicinity of the data (although see Hill (2006)). Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
  • 114. Summary and Conclusion Practical Pros and Cons The main drawback of our approach is that it is based on an assumption. The generalised CES is “not-quite-superlative”, and therefore we cannot just assume that the approximation to true preferences will be good in the vicinity of the data (although see Hill (2006)). Nonetheless it buys the ability to address the new-characteristics problem without any of the difficult econometric work involved in solving estimated hedonic demand equations for reservation prices. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 40 / 41
  • 115. Summary and Conclusion Practical Pros and Cons This is, we would argue, a practical plus. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
  • 116. Summary and Conclusion Practical Pros and Cons This is, we would argue, a practical plus. This is because NSI’s are never keen to adopt regression-based methods when producing their inflation data because of (entirely valid) worries over robustness and reproducibility. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
  • 117. Summary and Conclusion Practical Pros and Cons This is, we would argue, a practical plus. This is because NSI’s are never keen to adopt regression-based methods when producing their inflation data because of (entirely valid) worries over robustness and reproducibility. Extrapolation of estimated hedonic models/demand equations might therefore be a step too far. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41
  • 118. Summary and Conclusion Practical Pros and Cons This is, we would argue, a practical plus. This is because NSI’s are never keen to adopt regression-based methods when producing their inflation data because of (entirely valid) worries over robustness and reproducibility. Extrapolation of estimated hedonic models/demand equations might therefore be a step too far. Our method has the major advantage of not requiring NSI’s to do any econometric work over-and-above what they are doing already. Crawford & Neary (Oxford) New Characteristics RES-ONS: July 19, 2018 41 / 41