Discussion of Bosworth-Triplett on Transportation and Communication

1
Discussion of Bosworth-Triplett on Transportation
and Communication

• Robert J. Gordon
• Northwestern University and NBER
• Brookings Workshop on Productivity in Services
  Industries Trends and Measurement Issues,
• November 21, 2003

2
What the Authors Achieve in These Two Papers

• Extremely Clear and Valuable Explanation of BEA
  and BLS Measurement Methods
• BEA vs. BLS Coverage Differences Tied to SIC
  Codes
• Table 4 for Trucking Excellent, Underlines
  Complexity
• Table 5 for Airlines Likewise
• Discontinuities in Methods, Coverage, Deflation
• Whatever said later about Gordon (1992), B-T do
  the BLS vs. BEA comparison MUCH better
• Uniform Comparison of Labor Productivity Growth
  Rates, BEA GO, BEA VA, BLS

3
Beyond Accounting for BEA vs. BLS, focus on BEA
MFP Growth

• Uniform format to Link Labor Productivity to MFP
• Contribution of Capital per Worker
• Contribution of Intermediate Inputs
• Good Job of Highlighting Surprising Results,
  e.g.,
• Disappearance of MFP Growth for Airlines
• Unbelievably rapid productivity decline for
  Radio-TV

4
Losing the Forest for the Trees

• B-T Papers (both Transportation and
  Communication)
• More Serious in the Transportation Paper
• Lose Sight of the Role of Purchased Capital in
  Creating Productivity Growth
• They Use BEA data on Purchased Materials as if
  they were fact rather than Measurement Error
• B-T produce the best evidence yet that BEA VA vs.
  Intermediate is a fraud

5
My Discussants Perspective

• Point of Departure, my parallel paper on
  Transportation in the Griliches CRIW Volume (50
  printed pages)
• (First reaction B-T papers are too short, esp.
  Communications!)
• Natural Questions Arise
• Are Current Data for 1977-87 Overlap the Same?
• Substantive Changes in Productivity Behavior
  since 1987?
• Balance between Measurement and Substantive
  Analysis
• We come back to Substance, no time for other

6
Ive Got Three Main Points, Heres Big Point 1

• Why Should We Expect Any MFP Growth in
  Transportation?
• Why Should We Surprised if Measured MFP Growth
  Accelerates or Decelerates?
• The Transportation Industry buys Big Boxes and
  Engines to Propel the Boxes
• Then It Puts Stuff Inside the Boxes
• Transportation Industry does no RD, its all
  done by the makers of the Big Boxes
• GE Locomotives and Aircraft Engines
• Boeing Aircraft
• Cummins, Detroit Diesel truck engines

7
Thus the MFP Action for Transportation Must
Reflect the Quality of the Boxes and Engines

• MFP Changes Hide Errors in the Price Indexes for
  Capital Inputs in Transportation
• Simplest Example Airlines
• The History of MFP Growth in Airlines reflects
  unmeasured quality change in the boxes and
  engines
• Especially the transition to Jets

8
What is the Evidence?

• Rapid Measured MFP Growth in Airlines reflects
  unmeasured quality growth in airline capital
  prior to 1977
• Labor productivity growth reflects changing
  growth in aircraft quality, DIMINISHING RETURNS
• 1935-69 7.1
• 1969-78 5.7
• 1978-87 3.8 (B-T 3.6 Table 2)
• B-T 1987-2001 BEA GO 0.1

9
But What About Those Obvious Productivity
Improvements?

• Airlines have cut back on the use of purchased
  inputs (travel agents), and own-employees (res
  agents, CTOs)
• Yet every innovation that allows airlines to cut
  back on its own employees comes from a supplier
  of capital goods
• Web developers
• Kinetics, one of the companies that develops
  e-check-in kiosks

10
Overall Conclusion on B-T Treatment of Airlines

• They Miss Entirely the Role of Diminishing
  Returns in the Growth of Capital Quality
• Diminishing returns in capital quality have
  Reduced the Growth of Labor Productivity and MFP
• Subtlety Since 1980 Most Quality Improvements
  in Capital Take the Form of Better Fuel Economy
  and Longer Range (relevant only for international
  aviation)
• This All Should Come Out in Price Indexes for
  Aircraft, but B-T dont discuss this at all

11
Beyond Airlines, What Role Does Capital Play?

• Trucking
• Capital is the entire ball game
• Three types of capital improvements
• Size and fuel economy of engines
• Minor role of IT capital discussed by B-T
• But overwhelmingly the most important role of
  capital input was . . .
• GOVERNMENT-FINANCED HIGHWAY CAPITAL!!
• Long-live Barbaras Highway Capital!
• Explicitly taken into account in MFP calculations
  for airlines and trucking in Gordon (1992), not
  mentioned in B-T (2003)

12
Whats Left Besides Buying Capital?

• Transportation companies face a tradeoff between
  improving service and improving productivity
• Shorter check-in queues or faster baggage
  delivery requires
• More Employees
• More or better capital (e.g., baggage scanners)

13
Second Big Conclusion, B-T dont heed their own
advice

• The BEA Gross Output Series Should be used in all
  Analyses of Industry Productivity
• The BEA Value-added series are fatally flawed
• This means that none of the B-T tables should
  list either
• Labor Productivity based on VA
• Contribution of Intermediate Materials
• MFP based on VA

14
How Can I Reach Such an Extreme Negative
Conclusion on VA?

• Refer to B-T communications paper p. 5
• capital-type income reported to IRS on a company
  basis
• Assignment of incomes requires a conversion to
  establishment basis
• No good ways to make the conversion
• Likely to be significant inconsistencies between
  estimates of value added and gross output

15
The Killer Evidence

• Look at B-T Communications Paper, Figure 1
• No resemblance BEA purchased input shares to
  Census input share
• Census Share is Very Stable
• My Conclusion all these tables should be redone
  to assume a near-constant materials share
• Get your white-out and cut out two lines in B-T
  Tables contribution of intermediate and MFP
  based on VA

16
More Killer Evidence that We Should Bury the BEA
VA Program

• Look at Airlines, B-T Transportation Paper Table
  2
• What is the Contribution of Intermediate Inputs?
• 1977-87 per year 2.5
• 1987-95 per year -2.3
• How Could Price-Deflated Intermediate Inputs
  Increase 25, then Decline 19??

17
Before Leaving Airlines

• The B-T paper contains offhand comments about
  changes in airline output quality
• Fully treated in Gordon (1992)
• Increased seating density occurred 1975-1982
• No comments on
• Value of frequent-flyer awards
• Narrowbody TV, e-kiosks, web booking

18
How I Remember the Iraq War Mark I (Jan-Feb
1991)!!

• Really Annoyed by Critics who Said Deregulation
  required people to connect by circuitous routes
• Watched CNN, started counting
• Did a complete count of the entire US airline
  network, 1978 vs. 1989, routes and frequencies
• Concluded a huge increase in quality, many more
  new routes than dropped routes

19
Conclusion 3 BEA Backwardation

• BEAs huge advance in benchmark revision of
  industry data, dated January 1991
• Forced me to redo my entire paper midstream
  between May 1990 conference and 1992 publication
• Changed from output indexes based on flawed
  deflators (noted by B-T) to volume-based indexes
• B-T tell us as of 1992 BEA has moved backwards to
  flawed PPIs (see B-T Transportation Fig 3)

20
B-T Are on Top of Conclusion 3, they just bury
and dismiss it

• Look at B-T Treatment of Railroads in
  Transportation Paper
• The B-T discussion of railroad output includes an
  excellent criticism of the BEA use of the flawed
  PPI
• Long live the discrepancies between the BLS
  productivity program and the BEA at least the
  BLS eschews use of its own flawed PPI!

21
Railroads, What Are the Issues?

• Shift in the Composition of Output?
• Stark Comparison
• Boxcars of flight by unemployed in the 1930s
• Cattlecars of the holocaust
• Vs. todays specialized larger cars for coal,
  grain
• Union Pacific TV Commericals
• Containers are a form of outsourcing

22
Small Comments on Communications

• Analogy to Transportation Almost all the
  productivity growth is being produced by
  purchased capital input
• Just as there was no discussion of deflators for
  capital input in transportation, no discussion of
  deflators for capital input in broadcasting
• Good discussion and use of best recent research
  on deflators for communications capital

23
Caution regarding Broadcast Productivity

• My Back Yard
• Huge Investment in late 1990s in Extending
  Digital Cable that last mile
• Is that properly decomposed into current expense
  and capital expense
• I am paying huge monthly cable bills that are
  eventually going to amortize that investment
  (100 40)

24
Their Disappointing Conclusion

• We need (Trans, p. 22) closer integration of the
  two agencies work on industry-level output and
  productivity.
• How about an alternative verdict?
• We need to have BEA abandon the current
  methodology of its value-added program
• We should prohibit BEA from using PPI deflators
  where an alternative volume-based is available

25
The Final Broad Perspective vs. Where We were in
1992?

• Compared to that 1992 paper, are we better off?
• This paper contains critical indictments of BEA
  measurement methods but buries them rather than
  highlighting them
• Flawed VA-materials decomposition
• Flawed reliance on PPIs instead of volume
• Papers lack of overall recognition that
  productivity gains in these industries come from
  purchased capital, not own-innovations