Productivity and the Business Cycle: Evidence for Europe and the US (EACBN)

1
Productivity and the Business Cycle Evidence
for Europe and the US (EACBN)

• Discussion for
• Technology Shocks and Job Flows
• by Claudio Michelacci and David Lopez-Salido
• Sumru Altug
• November 29, 2005

2

• The purpose of this paper is to examine the
  short-run response of employment and output to
  investment-specific and neutral technological
  shocks. The paper combines a version of Solows
  growth model with a Mortenen-Pissarides search
  technology which leads to sluggish reallocation
  of workers in the labor market.
• Investment-specific technological advances make
  new capital equipment less expensive whereas
  neutral technological progress creates
  productivity gains even in the absence of capital
  investment. The paper identifies the impact of
  the different types of technological shocks
  through long-run restrictions in structural
  VARs.
• The investment-specific technological shock is
  identified from the fact that it is the unique
  driving force in the relative price of investment
  goods. By contrast, neutral technological
  progress is assumed to explain any remaining
  component of the trend in labor productivity.

3

• The paper is related to earlier findings by Gali
  regarding the impact of technology shocks on
  aggregate hours worked. In Galis framework, the
  sole technology shock leads to a decrease in the
  aggregate number of hours worked.
• By contrast, the authors find that
  investment-specific technological shocks lead to
  an expansion in employment, output, and
  investment whereas neutral technological shocks
  lead to jobless recoveries as the destruction of
  technologically obsolete jobs goes hand-in-hand
  with the creation of new technologically advanced
  ones.
• These results are derived through a structural
  VAR analysis and through the impulse responses
  for the model economy.

4
Is RBC analysis turning into fast food?

• The paper is, in general, carefully executed. It
  motivates the model through an initial VAR
  analysis, and also describes and solves the model
  with relative skill. Nevertheless, I couldnt
  help feeling a little dissatisfied with the whole
  exercise once I was done reading the paper. I
  felt as if I had eaten a Big Mac its certainly
  filling but one still hungers for real food
  afterwards.
• Why do I compare the paper to a Big Mac? Well,
  its as if it takes a ready-made beef patty and
  throws it on the grill and cooks its for a few
  minutes and then it packages the whole thing with
  a bit of lettuce and tomato and ketchup and
  mustard and serves the whole thing in a
  ready-made bun. I initially considered titling my
  discussion with the title of this slide. Let me
  describe in more detail my reasons for thinking
  this way.

5
Use fresh ingredients

• The paper has too many ready-made features the
  Solow growth model, the Mortensen-Pissarides
  search model, Blanchard-Quah type long-run
  restrictions on VARs, the Gali hours-technology
  result and more.
• Wheres the new model? Wheres the new
  application? Wheres the new data? Wheres the
  new simulation approach? Wheres the new
  econometric trick?

6
Dont engage in false advertising

• I have to admit that I dont like throwing a bit
  of micro evidence or institutional detail to
  motivate macro models. Its like putting a bit of
  lettuce in the Big Mac and making people think
  that they are eating healthy. In this paper, some
  of the most interesting tidbits of micro or real
  life evidence are in footnotes.
• What constitutes an investment-specific
  technological advance versus neutral
  technological progress is described in footnote
  3. Yet one might argue that how this distinction
  works in practice is really the cusp of the
  paper. If we are to believe that
  investment-specific versus neutral technological
  shocks really do have differential effects over
  the business cycle, then we would also like to
  understand much better how the introduction of
  software, more powerful computers or more
  efficient means of telecommunication and
  transportation differs from improvements in
  accounting techniques or in the organization of
  production, marketing and management control in
  a cyclical sense.
• Are the latter what we mean when we talk about
  neutral technology shocks? Or are these
  improvements just changes in trend productivity?
  If so, then how we distinguish investment-specific
  versus neutral technology shocks in practice?

7
Cook a filet mignon

• The study of Economics is now quite advanced that
  many of us are tired of consuming fast-and-ready
  answers to difficult questions. Footnote 4 of the
  paper quotes Gordons (1990) study which provides
  examples from different industries regarding the
  adoption of new technologies.
• The authors use these findings to argue that the
  adoption of new technologies may require the
  performance of some new tasks by workers. I was
  tantalized to know Which industries require
  which tasks? How do industries differ in their
  technology adoption? How does this affect the
  nature of worker-specific tasks?
• To answer these questions, the authors could have
  used sectoral data instead of aggregate data in
  their analysis. This would also provide a better
  indication of the importance of
  investment-specific versus neutral technological
  shocks over the business cycle.

8
If you want to eat Chinese food, go to a Chinese
restaurant

• Having expressed my dissatisfaction with the
  description of the shocks, I was equally
  dissatisfied to see references to Aghion and
  Howitt (1994), Mortensen and Pissarides (1998),
  Jovanovich and Lach (1989) and so on and as a way
  of motivating the theoretical model.
• These papers are examples of sophisticated
  theoretical analyses that provide deep insights
  into the search process and into technology
  adoption. Incorporating features from these
  models into the present growth model was a little
  bit like serving Chinese egg rolls in a fast-food
  restaurant.

9
Should we go hungry?

• By this time, you must be thinking that I am
  advising against any lunch. Thats not the case.
  Indeed, it looks like most of the insights of the
  model can be had from the decomposition of labor
  productivity provided in Section 2. That
  decomposition is the basis of the identification
  of the structural VAR and the resulting empirical
  results.
• It appears that there is an original and readable
  paper up to page 15. Its when we got to the
  model and especially the equilibrium conditions
  that I started to balk. I might be induced to
  partake of the numerical simulation results
  implied by the model but they have to be served
  up more quickly.
• Why not describe in words the implications of the
  model, and relegate much of Section 4 to an
  appendix?

10
Calibration

• Ive complained of quite a lot by this time so
  its not surprising that I might end by bringing
  up the issue of the calibration. There are too
  many parameters. Many of them are derived from
  plant-level studies. See, for example, Cooper and
  Haltiwanger (2002) or Baily et al (1992).
• Are we doing a disservice to these papers by
  selecting some parameter values and plucking them
  into an aggregative model?
• Have we really explained the aggregate
  fluctuations?

11
Conclusions

• This concludes my comments. Would I have written
  this paper? The answer is probably no.
• Should such papers be written? Yes but with more
  insight.
• Where is this research strategy leading us? We
  can always uncover more VAR results at the
  aggregate level but it will take a lot more
  convincing to link them to any plausible
  mechanism of the business cycle.

12
(No Transcript)