The Boskin Report vs. NAS

1
The Boskin Report vs. NAS At What Price The
Wild vs. the Mild

• Robert J. Gordon,
• Northwestern University and NBER
• CRIW, Cambridge MA, July 31, 2002

2
Puzzled about the NAS Report

• Invited to this panel, my first problem was
  obtaining a copy
• First puzzle Why was this report necessary?
• Was the Boskin Commission so inadequate?
• What about the Conference Board report?
• How many times does the CPI need to be evaluated
  within a 5-year period?

3
Stated Motivation

• We find out on p. 17, the BLS asked for it,
  commissioned it, paid for it
• Their charge exactly the same charge that the
  Senate gave the Boskin Commission
• Investigate conceptual, measurement, etc.
• Appropriate use for indexing Federal programs
• Cant stop myself from WWII analogies
• BLS Britain
• Boskin Luftwaffe
• NAS Americans riding to the rescue
• BLS Help! Weve been attacked, what should we
  actually do?
• They had already achieved much post-Boskin on
  their own
• What should remaining priorities be?

4
Similarities and Differences, Boskin vs. NAS

• Similarity same topic
• NAS less interested in substitution since CPI had
  already adopted Boskin proposals upper-level and
  lower-level substitution was at the core of the
  Boskin proposals.
• Formula (lower-level) bias had only been
  recognized shortly before the Boskin
  deliberations
• Similarity A committee
• Individual NAS chapters were primarily the work
  of particular people, just like individual
  sections of the Boskin report

5
Big Differences

• We were charged with coming up with an exact
  numerical estimate of the bias, at least we
  thought so and never debated it
• A terrific idea
• In retrospect, it created a central magnetic
  field everyone could disagree and tug the
  number down or up, but it was there
• Todays central bias number exactly what Boskin
  report said minus what the BLS has fixed
• Nobody had ever committed before
• No numbers about bias in the 1961 Stigler report
• Thats one reason the NAS report feels wimpy

6
The Biggest Boskin Innovation The Quality
Change Matrix

• Quality Change and New Products
• We took 27 industries, said we knew nothing
  relevant for about 9 of them, and the other 18 we
  came up with a precise number, what chutzpah!
• Most important point using related research
  to extrapolate bias from one industry to another
• We scribbled on the back of a very big envelope

7
Big Difference Boskin Lack of Review, Speed,
Spontaneity

• Boskin commissioned June 95, final report with
  hearings first week Dec. 96
• We were a creation of the Senate Finance
  Committee and especially of the
  soon-to-be-disgraced Bob Packwood and then the
  elegant Daniel Moynihan
• In between, a back-burner activity
• Perhaps 5 meetings, of which 2 or 3 were with the
  top BLS people

8
How decentralized it was!

• The Boskin Comm was built on trust
• No arguments, no bickering
• Dale said 1.1 implies 1 trillion in SS savings
  over 10 years
• Somehow our separate efforts came up with the
  1.1 bias number, but Zvi and I didnt question
  the substitution part, and Mike/Dale/Ellen didnt
  question the 0.6 quality change part

9
Boskin Stories

• Kathy Abraham was a hero
• We were the barbarians at the gate, the Luftwaffe
  over London
• She handled us masterfully
• Much praise also to John Greenlees and his staff
• Those cozy meetings in the conf room of the
  Senate Finance Committee
• Another WW example Christmas eve on the Western
  front
• Our reward was not financial, it was Moynihan!
• The day of Packwoods demise, viewed from
  Moynihans retreat (Gentlemen, . . . )
• Lunch in the Senate, the famous navy bean soup

10
When was the Boskin Report actually written?

• Four main parts
• What the BLS actually did
• Implications of various CPI-X formulae for the
  Federal budget 10 years into the future
• Substitution issues lower level, formula bias,
  upper level
• Quality change and new products

11
The Most Controversial Part Quality Change and
New Products

• Press Conference was sked for Dec 4
• Quality Change/ New Products written in the two
  weeks before November 26, Fed-exed to ZG the day
  before Thanksgiving, in a world pre-email
  attachments ( Nov 28)
• Billions slipping off the keyboard
• Two hour phone call on Sat Nov 30, Cambridge to
  JAX (who paid?)
• When in doubt, cut the number by half

12
Quality Change Section

• There could be no bigger difference from the NAS
  report, in lack of review
• Once off the phone with ZG, the quality change
  section was in Fed Ex to Mike Boskin
• Mike did a heroic job in merging these sections
  and getting a coherent report to Washington two
  days later.
• Reviews? What reviews? We knew the reviews
  would come later and they came, in spades

13
Once Released, It Became Politics

• How many TV cameras that day?
• We were going to revolutionize Federal Finance
• The big question became not the details of the
  report but rather
• CPI minus X (where X was the 1.1 number)
• CPI minus AX (where A was the fudge factor)
• Eventually scuttled by AARP and other lobbies

14
So What Happened After Publication?

• Unlike NAS, which was heavily reviewed before
  publication, we were reviewed after publication
• Much of Boskin report is a first draft
• Brent Moulton and Karen Moses gave us a masterful
  beating up (BPEA 1997, no. 1)
• But they approved of the matrix framework as a
  way of organizing ideas, no one had ever done
  that before
• Their nit-picking peeled away maybe 0.15 of our
  0.6 quality change bias, but we still had new
  products in reserve, for which we had not
  developed any estimate except the presumption of
  the direction of bias

15
Recommended Scope of CPI

• Limit it to Private Goods and Services
• BLS should not be drawn in to measure impact of
  environmental changes, benefits of anti-pollution
  legislation, public goods, increases in life
  expectancy
• Agree The Boskin Comm speculated at end about
  our report about how these issues might affect
  our bias estimates without recommending that the
  BLS do anything about them

16
Some Topics Should be left for Academic Research

• Over time, effects of environmental decay and
  pro-environment legislation
• Across countries, issues in comparing standards
  of living
• Europe vs. US. Air conditioning, heating, energy
  use, auto use vs. public transport, low density
  of U. S. metro areas, vs. undeniable U. S.
  advantage of larger home and lots sizes

17
Downplay disaggregation among consumers

• Agree, same as Boskin
• No need for CPI to create special-purpose price
  indexes for the elderly
• Not to mention males, females, ethnics,
  teenagers, or economics professors

18
Differences in Emphasis

• In Boskin, much more beating up on CPI about
  upper-level and lower-level substitution issues
• But CPI accepted that right away and acted with
  amazing speed to change at both the upper and
  lower levels
• Some of this had been planned before Boskin
• Not much left for NAS to criticize

19
Whatever Happened to the Second Research-Based
CPI?

• The CPI can never be revised, we all agree
• We need an alternative CPI that continuously
  incorporates results from historical studies,
  goingn back, at least in principle, to 1914
• Current BLS CPI-RS is a good start, but needs to
  have differences between CPI-RS and CPI-U and
  CPI-U-X1 split apart in a regular table available
  to anyone
• Remarkably little attention to alternative CPI in
  NAS report
• Mentioned in passing on p. 8, but not as a
  numbered recommendation, no emphasis on its
  potential for historical research

20
COGI vs. COLI

• Need minimal use of COLI theory to perform
  practical comparisons Agree
• Discussion muddled
• COGI vs. COLI differ in assumptions about
  substitutions and in weighting schemes
• No difference in the difficulties posed by
  quality change and new goods
• Good suggestion not in Boskin Adopt advance
  estimate of the superlative index

21
Quality Change

• Both reports share the BLS emphasis on
  incremental improvements going forward
• Lack of attention to potential for improving our
  measures of quality change going backward into
  history
• Implicitly, let the academics do it

22
NAS on Hedonics

• To be skeptical is not novel
• Triplett automobiles may be too complicated
  for the hedonic technique.
• Hedonics best for simple products, PCs are a
  great application
• PC memory and speed have standard measures, new
  features (CD drives etc.) can be dummied out
• Picture quality of TVs and Sound quality of
  stereo audio more problematic
• Caution exaggerated, Rec 4-3 (p. 7) wimpy

23
Defects of Matched Model Indexes

• Not enough emphasis either in Boskin or NAS
• Price declines missed when they occur with new
  model introductions (PCs)
• Price increases missed when they occur with new
  model introductions (womens apparel)

24
Independent Advisory Board?

• Recommendation 4-8 (p. 7)
• econometricians, statisticians, index experts,
  marketing specialists, product engineers
• Would have to hire a ballroom to get them all
  together into one room
• Unwieldy and unnecessary. The BLS is doing just
  fine, implementing hedonics incrementally on its
  own

25
New Goods?

• NAS says do nothing
• Thats where a retrospective research-based index
  could make a major improvement
• Example VCR introduced into the market at 1200
  in 1978, by 1987 price had fallen to 200
• Introduced into the CPI in 1987
• Low-hanging fruit
• Many similar examples, e.g., air conditioners
  1951 vs. 1964

26
Outlet Substitution?

• NAS Do nothing other than conduct research
  (rec 5-2, p. 9)
• Data are available now on the market share of
  discount stores going back at least two decades
• Another issue that a research-based historical
  index can address

27
End on a High Note

• Medical care, Chapter 6
• Barely treated by Boskin
• Except to cite a few studies Cutler, Shapiro,
  etc.
• I agree with the major recommendations
• Develop a new medical care total expenditure
  index (expanded scope medical CPI)
• Base it on a substantial number of treatment
  episodes

28
Data Collection

• Excellent discussion of scanner data, NAS largely
  comes out pro-scanner
• But did not pick up on Boskin distinction between
  local data (rents, vegetables) and national
  data (consumer durables)
• No need to collect so much data on a local basis
  for nationally sold durable goods

29
Finally, CPI vs. PCE Deflator

• Good discussion of possibility of merging
  upper-level weights
• But lack of discussion of ongoing divergence
  between inflation rates of CPI and PCE deflator
• As CPI methods have improved and weights updated
  faster, should be convergence
• But instead, more divergent than ever
• Should be a permanent ongoing table to provide a
  decomposition of reasons