Equality, Quality, Big data Flashcards
Weights?
- Given price data – how to combine into index?
- Unweighted averages of various sorts - Carli, Dutot, Fleetwood, Shukburgh-Evelyn : prices 1050-1800
- Laspeyres and Paasche – need quantities
Expenditure weights?
- Emerged in early 20th C. – linked to measures of cost of living for working class
- First UK comprehensive expenditure survey – 1904 survey of HH expenditure
- Food items important – inc quality and p info
- 1913 – enquiry into working class rents and retail p
- Used to adjust wages to changes in cost of living during WW1
Living cost and food survey?
- Address based sample of HH
- Primary Sampling Unit is post code sector – total of 638 per year – 12 monthly waves
- Within each PSU, 18 addresses are chosen (interviewer quota), PSU’s are stratified sample using –
- Regions – 12
- Socio-economic group – 4 bands according to proportion of HH whose head was in professional/managerial occupation
- Car ownership – PSU sorted in order of proportion of HH w no car
- Unit of sampling is a HH – HH comprises one or more persons (not necessarily related) living at same address sharing cooking facilities and a living/sitting/dining room
- About 12,000 in a sample
- Main aim of LCF to find out expenditures by each HH for each category (COICOP, group, class or sub-class)
- Face to face interview to collect regular expenditure (utilities such as mortgage, insurance etc.), followed by self completion diary recording expenditures for HH for 2 week period
- Survey doesn’t measure prices paid by HH or quantities, except in cases where one unit purchased
- Equivalised HH – adjust for HH size and composition
Latest reprot - family spending in UK - April 2018 to March 2019
- Avg weekly HH expenditure in UK £585.60 – highest since financial yr ending 2005 after adjusting for inflation
- Transport was category w highest avg weekly spend of 14% off HH avg total weekly
- Highest in London and South East, spending in North East the lowest
Plutocratic v Democratic weights?
- Long debate over how should weight expenditure shares
- Plutocratic – add up expenditures across all HH, then divide into exp shares – in effect weights HH by their expenditure, rich more important – spendthrifts more than the thrifty savers
- Democratic – divide into exp shares at HH level, average across HH – each HH has same weight – each person has smaller weight the bigger their HH
Can construct CPIH using either but doesn’t seem to make too much difference
What is a household?
The ONS defines it as
“A household is one person living alone, or a group of people (not necessarily related) living at the same address who share cooking facilities and share a living room, sitting room or dining area”.
To get the democratic measure, you can give the household a weight depending on the number of people (adults?), or the equivalised index?
“Equivalisation is a standard methodology that adjusts household income to account for the different financial resource requirements of different household types. Household size is an important factor to consider because larger households usually need a higher income than smaller households to achieve a comparable standard of living. The composition of a household also affects resource needs, for example, living costs for adults are normally higher than for children. After equivalisation has been applied, households with the same equivalised income can be said to have a comparable standard of living.
Expenditure shares by category of spending, by equivalised disposable income decile?
COICOP 3, 7, 9 and 11 are luxuries: the expenditure shares increase with income.(Clothing and Footwear; Transport; Recreation Sport and Culture; Restaurants and Hotels)
Rest decrease, except for 10 which is U shaped (Education)
Differences with CPI - uses UK domestic population?
using a measure of direct payments to calculate OOH
using a measure of direct payments to calculate the cost of education (for example, including tuition fees as they are paid for via student loan repayments, rather than including the upfront cost)
using a measure of direct payments to calculate items that are paid for in advance (for example, package holidays, airfares and cultural events)
including the full cost of insurance premiums without accounting for the value of insurance claims received (or deducting expenditure using insurance claims received from the relevant items, instead of insurance itself; for example, expenditure on new cars would be adjusted to exclude any expenditure on new cars that resulted from an insurance claim to avoid double-counting this expenditure)
including the cost of interest on all debt (for example, mortgage interest and interest on credit cards and loans)
including the cost of goods bought second-hand without taking into account goods sold (such that household-to-household transactions are included as a cost)
including items that are seen as saving in traditional consumer price index methodology (such as capital housing costs, savings and pension contributions)
How do you measure changes in quality?
In the SR ignore it, it doesn’t matter - inflation primarily a matter for contemporary economic policy
But in LR a lot of little changes accumulate - inflation measures are used for indexation of pensions and benefits - a retiree may be relying on a pension for 30 years or more
Am I earning more or less than I was 30 years ago, or 10 years ago? Very relevant fo r understanding the observation that median hourly real wages have not increased in theUS since the 70s (or in the UK since 2008).
Is the UK productivity slowdown simply a reflection of the fact that we are not measuring quality correctly?
Official Line?
Within the item specification, the specific products chosen are from amongst the most popular (in that location and/or shop). When there is a change of product (for example, when you change shop the previous brand is not available), the collector is supposed to choose a replacement product that is of the same quality.
If a product is temporarily out of stock, the collector chooses a substitute (using codesN,C,T,M) on the basis of “similar quality and quantity”.
When there is a new product with a new price, you can do one of several things, including
(a) Treat it as a pure price change. Assume that the quality is the same.
(b) Treat it as a pure quality change. Assume the price difference reflects difference inequality: “true price” unchanged (price per unit quality).
(c) Where there is a only a change in “size”, use unit price to link the old and new(shrinkflation)
Really very simple: ONS uses options (a) and (c). No explicit account is taken for quality changes. Except for a limited range of goods
Hedonics?
Dates back to 1939, Andrew Court: used to measure quality change in automobiles (Ford motor company in US). Used in Housing indices in UK and elsewhere.
CPI3 and CPIH: Used for smartphones, PCs, Tablets.
Method: You take a list of “value determining characteristics” (for PCs, processor speed, memory, hard drive etc.). Each month, you run a pricing regression (a log-linear regression of actual price on these characteristics). This enables you to obtain a“predicted price”, which is a measure the quality. Two PCs with the same predicted price are judged to have the same quality
Hedonics - how to use?
When you calculate price relatives with respect to previous January, you adjust for quality change:
For example, the July price relative for PCs is the ratio of the July price of the PC, divided by the quality adjusted value of the base price.
Quality adjusted base price is actual price multiplied by the ratio of the predicted price of the July PC divided by the predicted price of the January PC (using the July hedonic estimates). This increases the nominal price of the base year, reducing the price relative(and hence inflation).
Hedonic estimation is very labour and data intensive, so only used where there is fast changing quality variation. Could be used for lots of other products (washing machines, cars), but is not done because too expensive.
In the US, the Bureau of Labour Statistics (BLS) which computes the US CPI uses hedonics much more
Services?
Notoriously difficult to measure as an output and to capture changes in quality
Are funeral directors, priests, doctors more efficient than 20 years ago?
Services - in practice?
Item descriptions fairly general - if we look at national income accounts, there are some examples of quality adjustment but quite rare
Most common to measure quality by their ‘output’ or outcome, or input (costs of provision)
In national accounts, an adjustment made for output of secondary education in terms of increasing GCSE grades - output assumed to increase at 0.25% per annum
Lots of suggestions (health outcomes, survival rates) nothing much done
