Riccardo Fiorito

1
Riccardo Fiorito Università di Siena Why Macro
Labor Supply Is More Elastic Than Micro Theory
and Policy Implications Brown Bag Lunch
Meetings Dipartimento del Tesoro, Ministero
dellEconomia, Roma, 18 Aprile 2008
2
Background and Motivation

• This presentation aims at showing that micro and
  macro labor supply elasticities dont have to be
  the same (Fiorito-Zanella, 2008) as it is
  typically claimed by critics of the RBC model.
• While this is well known in the theoretical
  literature (Prescott, 2006), we show that the
  same result can be obtained empirically by using
  an appropriate aggregation procedure.

3
Background and Motivation

• Estimating and understanding this elasticity is
  not simply an econometric issue since
• It involves deep differences in the way of
  addressing macro-economic modeling.
• It involves also several fiscal policy
  implications as far as employment and
  participation decisions are concerned.

4
Macro Modeling A Quick Refresher

• Macro modeling started after WW2 using
• NIPA definitions (Meade-Stone, 1943 Kuznets,
  1946 SNA, 1953 etc.)
• Econometric tools (Cowles Commission) for
  estimating parameters inside large (non-linear)
  models to be solved (L. Klein, 1950).
• A macroeconomic paradigm based on Keynesian
  income determination (IS/LM), given fiscal and
  monetary interventions.

5
Reasons for Crisis

• Theory
• The Neo-classical growth model was not used in
  macroeconomics ? Business Cycles seen as
  deviations from an exogenous trend path.
• Macroeconomics did not consider individual
  choices.
• Taxation Lump-sum Taxation (? Expenditure only)
  rather than distortionary taxation affecting
  labor, spending and production choices.
• Policy confined to stabilization policies and
  unable to face supply shocks such as labor and
  oil shocks in the 70s and later.

6
Major Remedies

• VAR None or minimal identification (Sims, 1980,
  1986) estimation only matters.
• GMM univariate estimates More identification is
  needed (Hansen-Sargent, 1982) to estimate
  theory-driven parameters.
• Calibration Prescott (1986), Kydland-Prescott
  1982) Deep and other parameters can be obtained
  by micro estimates, NIPA definitions, theoretical
  suggestions nothing has to be estimated!

7
Calibration and RBC

• In principle, there is no reason that calibration
  applies to RBC only it is a wider GE methodology
  combining NIPA, stylized facts and sound
  micro-econometric evidence..
• Namely, microeconometric evidence could be used
  since
• A fundamental thesis of this line of inquiry
  is that the measures obtained from aggregate
  series and those from indivifual panel data must
  be consistent. After all, the former are just the
  aggregates of the latter
• (Prescott, 1986, p.22)
• While TFP persistence is widely recognized, labor
  supply elasticity is much more controversial
• Why so important?
• wages reflect labor productivity and affect labor
  input which enters the production function with a
  larger share that capital does..

8
Two Basic Parameters in the RBC model

• Technology is the driving force in the economy ?
  Impulses
• Intertemporal substitution of labor ? Propagation
  mechanism
• Empirics
• the TFP persistence is widely recognized
• the labor supply elasticity is much more
  controversial.
• Microeconomic studies always report a small
  elasticity (Men 0 - 0.2 Women 0 - 1)
  Blundell-Macurdy (1999) .
• Macroeconomic evidence mixed and less widespread
  because statistically weaker time-series never
  are long enough!
• Favourable cases Lucas-Rapping (1969)
  Alogoskoufis (1987)
• Non-favourable Hall (1980) Mankiw-Rothemberg-Sum
  mers (1985).

9
First View How Reconciling Calibrated and
Estimated Parameters?

• To account for all business cycle facts in the
  US, the RBC model requires a much larger labor
  supply elasticity than it is typically estimated
  in the micro literature.
• Recognizing this gap produced a possible
  inconsistency or else an incentive to extend the
  benchmark RBC model.

10
Reject or Extend the Benchmark RBC model?

• Reject (Summers 86, Mankiw 89)
• Extend Nonseparable leisure (Kydland-Prescott
  82)
• Indivisible labor (Hansen 85, Rogerson, 88)
• Home-production (Benhabib-Rogerson-Wright 91)
• Gov.t consumption (Christiano-Eichenbaum, 1992)
• Heterogeneous reservation wages (Chang-Kim 06)
• Human capital accumulation (Imai-Keane, 04)
• Nonlinearities between labor services and hours
  (Rogerson-Wallenius, 07)

11
A Second View

• Individual and aggregate labor supply
  elasticities are conceptually different
  (Prescott, 2006).
• Aggregate elasticity reflects technology too.
• Thus, there is no inconsistency if micro and
  macro elasticity differ.

12
Different Measures or Different Concepts?

• Individual elasticity change in individual
  labor supply (hours per worker) associated with a
  1 change in the after-tax real wage, holding
  wealth constant (Frisch)
• h ? (1-tau)w/p (intensive margin)

13
Aggregate Elasticity

• Aggregate elasticity change in aggregate labor
  supply associated with a 1 change in the
  after-tax real wage, holding wealth constant
  (Frisch) H ? (1-tau)W/P
• Aggregate elasticity is the product of the
  average hours per worker (h, intensive margin)
  times the aggregate employment stock N (extensive
  margin) i.e., it is a product of two different
  variables, not simply the aggregation of a single
  variable (intensive margin)
• Ht average( ht ) Nt

14
Intensive vs. Extensive margin
Hansen (1985)
Suppose we estimate individual and aggregate
elasticities as follows ?lnhit const.
e?lnwit uit , (Micro regression) ?lnHt
Const. E?lnwt Ut , (Macro regression).
15
Intensive and Extensive Margin
Then the elasticities are
So the aggregate elasticity is the sum of the
intensive margin and the extensive margin.
16
Our Contribution (Fiorito-Zanella, 2008)

• We derive a standard, life-cycle, labor supply
  equation (Macurdy, 1981) which is aggregated via
  home production. The latter accounts for changes
  in the extensive margin.
• Empirically, we use the PSID data to aggregate,
  each year, the individual units entering the
  labor supply equation.
• Since the micro (panel) and the macro labor
  supply (time series) are estimated exactly in the
  same way, we interpret differences in the
  estimated Frisch elasticity as reflecting
  aggregation only.
• In the labor supply case, aggregation means the
  product of the intensive and extensive margins,
  i.e. a different variable from micro.
• Aggregate labor supply is conceptually different
  from the individual one thus, there is no
  inconsistency in finding - or assuming - that
  aggregate labor supply is much more elastic.

17
Data

• The PSID is long enough (1968-2005) to apply our
  procedure. We use, however, all of the annual
  waves (1968-1997) to avoid interpolation of the
  missing years.
• Disadvantages no info on assets, tax rates, real
  interest rate and other important variables.
• Need to control for r anomalous years using
  appropriate dummies.
• How do our artificial series compares with the
  real ones?

18
Data PSID vs. USA
19
Data PSID vs. USA
20
The Underlying Model

• An economy with a single consumption good that
  can be provided by market or home-production.
  Labor is the only input, tax rate constant.
  Profit maximization with CRS technology implies

21
The Model Who Works on the Market?

• Obviously, in equilibrium
• So is individual is reservation wage,

22
The Model NFOCs

• Under CRRA specification,
• NFOCs of individuals who work on the market can
  be written as

23
The Model Deriving the Micro Regression

• Using standard procedures (Blundell and MaCurdy,
  1999), the Euler equation for labor supply is
• This is the micro regression where 1/? is the
  labor supply elasticity.

24
The Model Deriving the Macro regression

• We obtain the macroregression aggregating the
  first-order conditions and using the home
  production as shown in the paper
• Individual elasticity is
• Aggregate elasticity is
• The extensive margin is reflected by zt.
• For feasibility reasons we treat zt as constant
  since employment does not change too much.

25
Implementation

• We have two structural equations
• We do not correct for selection in order to
  precisely isolate intensive and extensive
  margins. However, when correcting for the
  selection bias, the micro estimate change is
  negligible.
• We will also add non-labor income as control in
  both.
• Consistency requires instruments. RE impose using
  lags because of econometric exogeneity. Lags of
  levels work better than lags of first differences.

26
Overall Results

• We find an individual elasticity of about 0.1, a
  low value that is standard in the literature.
• We find an aggregate elasticity of about 0.9, a
  much larger value not far from available
  calibration studies.
• 75 of the difference is due to the extensive
  margin i.e. participation/employment.
• Main conclusion in the labor supply case,
  microeconometric parameters are not a good source
  for calibrating aggregate models

27
Results 1
28
Results 2 decomposition of aggregate elasticity
29
Conclusions

• We do not aim at providing a new estimates of the
  micro regression we evaluate only the effect of
  aggregation. However, our micro elasticity is
  standard.
• When the individual and the aggregate Frisch
  elasticities are consistently estimated, the
  former is magnified by a factor of about 9.
• About 75 of this effect is due to the movements
  along the extensive margin.
• In this case, the micro parameter is not a good
  source for calibrating the aggregate because
  micro and macro labor supply are different
  concepts.

30
Policy implications

• While we cannot include tax rates, our results
  have also fiscal implications if tax rates dont
  change too much. Combining previous studies
  (Fiorito-Padrini, 2001) with stylized facts
  evidence, the ratio between the deviations from
  the smooth trend of the effective labor tax rate
  is always smaller than the corresponding
  volatility in real GDP the volatility ratio
  ranges for the G7 from about .5 for Canada and
  France to about .3 for the US, Italy and Germany.

31
Further Policy Implications..

• If the aggregate elasticity is close to 1 and if
  women elasticity is surely higher, reducing tax
  rates can have a significant effect on labor
  supply and, eventually, on the overall growth
  rate.
• This has several implications for government
  spending, including the incentive to provide
  services helping female participation. Likewise,
  transfers serving the opposite purpose should be
  reduced.
• Since the extensive margin dominates the
  intensive one, current proposals for de-taxing
  overtime work, seem to be less effective than
  general tax rate cuts acting on the most
  important, extensive, margin.