Legitimizing Concepts in Economics: An EvidenceBased Point of View

1
Legitimizing Concepts in Economics An
Evidence-Based Point of View

• Julian Reiss
• Departamento de Lógica y Filosofía de Ciencia
• Universidad Complutense de Madrid
• and
• Centre for Philosophy of Natural and Social
  Science
• London School of Economics

2
Overview

• Some general (and old) stuff on Evidence-Based
  Economics (EBE)
• Including a new slide 3 Distinctive Aspects
• Then a long case study about a hugely important
  concept the cost of living
• At the end, conclusions

3
The Ideal of EBE

• Evidence-Based Economicsis the conscientious,
  explicit and judicious use of current best
  evidence in making economic policy decisions
  about the welfare of societies.
• The practice of evidence-based economics means
  integrating
• individual socio-political expertise with
• the best available external evidence from
  systematic research. (Paraphrased from Sackett et
  al. 1996, p. 71)

4
Three Distinctive Aspects

• Fundamental orientation is political/practical
  (e.g. Clive Granger the aim of economics is to
  prepare policy decisions)
• Assumes a minimal level of objectivity (evidence
  is able to decide between competing ideas)
• Application matters (e.g., what is a good account
  may depend on what is doable)

5
What is EBE?II

• To urge that economic claims ought to be based on
  the best available evidence in neither empty
  nor trivial
• As we shall see, economic claims are often based
  on many other things but the best evidence
• Dogma
• Myth
• Theory
• Inapplicable or otherwise faulty methods
• Expert Judgement
• Etc.
• To say that a given claim is based on the best
  available evidence is to say that all possible
  (known) sources of error have been eliminated in
  the process of establishing the claim

6
A Case Study The (US) CPI

• Question How do we measure the change in the
  level of consumer prices?
• Huge debate in the US in the late 1990s. Reason
  Alan Greenspan claims that the CPI overstates
  true consumer price inflation by about 1 per
  annum
• Fact Federal debt would be almost 700 billion
  lower over a decade if the CPI was reduced by
  1.1 p.a.

7
The CPI Significance

• The CPI is of enormous practical and epistemic
  significance
• Practical
• Indexation of federal pensions (e.g. veterans)
• Indexation of social security programmes
• Indexation of federal debt
• Indexation of private contracts
• Indexation of tax brackets
• Epistemic
• Almost all empirical economic analysis is
  conducted in real terms, which means that
  variables are deflated by a price index

8
The CPI Measurement Matters

• Robert Gordon (1993)
• Price measurement matters. This is particularly
  true for errors in the rate of inflation over
  periods of three months to two years. In the
  American case, the late 1970s and early 1980s
  provided a textbook case of the follies that
  occur when policymakers rely on flawed price
  indexes. ... Not only were millions thrown out
  of work in 1980 and again in 1982, but the
  distribution of income was permanently distorted
  as millions of others receiving social security
  benefits escalated by the CPI enjoyed a windfall
  gain, since their indexed payments went up by
  more than the true inflation rate. The
  unnecessary extra benefit payments caused the
  government deficit to increase, and higher
  indexed union wages put more pressure on firms to
  raise prices and continue the inflation.

9
The CPI Indexing Payments

• Of all various uses of the CPI, here I focus on
  indexation
• Nixons Special Message to the Congress
  (25.9.1969)
• The impact of an inflation now in its fourth
  year has undermined the value of every Social
  Security check and requires that we once again
  increase the benefits to help those among the
  most severely victimized by the rising cost of
  living.
• I request that the Congress remedy the real
  losses to those who now receive Social Security
  benefits by increasing payments by 10 per cent.
• Beyond that step to set right today's inequity,
  I propose that the Congress make certain once and
  for all that the retired, the disabled and the
  dependent never again bear the brunt of
  inflation. The way to prevent future unfairness
  is to attach the benefit schedule to the cost of
  living.
• This will instill new security in Social
  Security. This will provide peace of mind to
  those concerned with their retirement years, and
  to their dependents.
• By acting to raise benefits now to meet the rise
  in the cost of living, we keep faith with today's
  recipients. By acting to make future benefit
  raises automatic with rises in the cost of
  living, we remove questions about future years
  we do much to remove this system from biennial
  politics and we make fair treatment of
  beneficiaries a matter of certainty rather than a
  matter of hope.
• i.e., in order to substitute expert judgement
  with mechanical objectivity

10
The CPI Reactions of the SFC

• After concerns were raised about the possibility
  that the CPI overstates inflation, a number of
  economists were asked to give testimony about the
  accuracy of the CPI (to the Senate Finance
  Committee)
• Then a commission was formed to study the CPI
  (interestingly, only those economists were asked
  to join the committee who in their testimony had
  claimed that the CPI overstates inflation)
• The Boskin Commission investigated without new
  research and reported on December 4, 1996 that
  the CPI overstates inflation by 0.8-1.6

11
The CPI Measuring Prices 101

• Basic idea track the cost for a fixed basket of
  goods through time (no economic theory
  whatsoever)
• US BLS collects the prices of 71,000 goods
  (called entry level items) in 22,000 retail
  units (plus information on rents and housing)
• Prices are aggregated into 207 item strata (e.g.,
  Apples, Physicians Services), which in turn
  are aggregated into higher-level series, and
  eventually into the CPI, using the modified
  Laspeyres (i.e., base-period weighted) formula
• Pit/Pil ?jqjbpjt/?jqjbpjl, e.g.
• Papples,1999/Papples,1998 ?jqj1995pj1999/?jqj19
  95pj1998
• Expenditure weights stem from the Consumer
  Expenditure Survey, conducted roughly every 10
  years

12
CPI The Boskin Report

• So whats wrong with this practice? Boskin
  commission everything
• Main flaw the CPI isnt a COLI
• That means the aggregation formula should be
  substituted by superlative indices
• That also means that new goods, goods whose
  quality has changed and new distribution channels
  should be treated differently
• BLS should establish a mechanism to bring in
  outside expertise
• Congress should pass legislation to address
  overindexing issue

13
CPI Whats a COLI?

• Idea tracking not of a basket of goods but of a
  fixed level of utility
• I.e., introduce an abstract quantity in order to
  make heterogeneous things commensurable
• Hence the Laspeyres index provides an upper
  limit to a true COLI
• You can also prove that under certain (in fact,
  implausible) conditions, the Paasche index
  provides a lower limit to a COLI
• A superlative index is one that (under certain
  conditions) approximates a COLI without
  information on the actual preferences

14
CPI Quality Changes

• The COLI framework, too, makes heterogeneous
  goods commensurable a good is a mere utility
  provider
• However, since utility cannot be observed (and
  measured only under unrealistic conditions), the
  effect of a quality change on utility must be
  imputed
• The BLS has a variety of mechanisms e.g.,
  mark-up method, link method, hedonic method
• The Boskin commission is effectively saying that
  the BLS methods underestimate the real impact of
  the quality changes

15
So Does the CPI Overstate Inflation?

• New Outlet Bias (0.1)- Boskin 30 (share of
  CPI) 0.7 (annual switching) 10 (quality
  adjusted price difference)- under a different
  interpretation consumers, because of declining
  real incomes, are forced to buy at discounters
• New goods and quality bias (0.6)
• ? far from being based on the best available
  evidence, the Boskin findings are based on
  thought experiments, inadmissible extrapolations
  from previous research and, in particular, theory

16
The Return of Expert Judgement

• Given how the world is, there is NO POSSIBLE
  EVIDENCE that can determine the true impact of
  quality changes on consumers utility
• Usually, new goods replace old ones
• Asking consumers what they would prefer were old
  goods still available may be an unreliable
  indicator of their preferences
• Actual (say, revealed) preferences may not
  coincide with reflective preferences
• So is the fate of benefit recipients necessarily
  in the hands of experts?

17
Should the CPI Be a COLI?

• The recent history of US price measurement
  provides an ideal case study for social
  constructivists evidence hasnt played a role
  either in switching from the price index concept
  to a COLI nor in determining quality changes,
  these are socially driven processes (e.g. a price
  index is more readily measurable while a COLI is
  a theoretically clearer concept)
• Alternatively, we can conclude so much worse (in
  this case) for the COLI
• However, that would be overly positivist (it
  doesnt exist unless we can measure it)
• Instead, the IN purpose should be the most
  fundamental criterion for evaluation
• The purpose of indexation does not (pace almost
  univocal agreement of the economics profession)
  determine that the CPI must be a COLI, quite to
  the contrary

18
An Alternative Take on Quality Changes

• There are three kinds of quality changes
• Repackaging more of the same
• Mere consumer teasers purely marketing-driven
  changes
• Genuine improvement
• If the purpose of the CPI is to measure the
  extent to which benefits recipients (etc.) are to
  be insulated from increases in the cost of
  living, we can argue that neither of the three
  kinds of change should be used to lower the CPI

19
Some Conclusions About the CPI

• First and foremost, the question what exactly is
  meant to be protected by indexing should a
  political decision, not one to be taken by
  economists
• Protecting the cost of living might mean
• Cost of meeting basic requirements (food,
  clothing, shelter)
• Cost of maintaining ones usual standard
• Cost of purchasing goods which are included in an
  accepted standard level of consumption
• Here are a few suggestions
• Rawls index measures the cost of obtaining
  basic needs
• Nussbaum index measures the cost of obtaining
  objective means for a flourishing life
• Democratic weighting
• Separate indices for as many subgroups as needed
  (but note trade-off between cost of additional
  research and just outcomes)
• In the proposed indices quality changes will play
  an attenuated role
• Conduct research on the degree to which people
  substitute
• However, because people differ, and because the
  world is evolving, any such index will
  necessarily be a somewhat fuzzy quantity

20
Some Conclusions About Economic Index Numbers

• There is an old approach to index number theory
  which says that an ideal formula must pass a
  number of (mathematical) tests
• Proportionality test
• Circularity test
• Etc.
• However, at least as important as satisfying
  these mathematical properties is satisfying a
  number of crucial material tests
• Utility test does the concept behind the IN
  serve the IN purpose?
• Measurement test are the associated measurement
  procedures consistent with the concept behind the
  IN?
• Dynamics test does the IN adequately deal with
  change?
• Robustness test are IN results robust enough to
  plausible specification changes (where enough
  is determined by the IN purpose)
• (Intuitions test) Are the IN results roughly
  consistent with our intuitions about the
  behaviour of the quantity of interest