Modelling labour market at the MNB Zolt

1
Modelling labour market at the MNB Zoltán M.
JakabSzirák2006. Nov. 4
2
Outline

• Macro modelling and labour markets, is it useful?
• Interest again focused on it among modellers
• Price and wage persistence
• Inflation fcast error analysis at the MNB, what
  are the main findings for the labour market
• Description of labour market models at the MNB
• NEM model
• Expert system in a nutshell
• new DSGE model
• Challanges to our models
• Capital/labour substituion
• Effective labour supply
• Substitution between public and private labour

3
Macro modelling and labour markets large focus
again

• Take a simple (hybrid) New-Keynesian Phillips
  curve common in the literature
• s denotes sectors, ? and ?? are Calvo-probability
  and share of indexation, respectively
• Emprical fact slow response of price inflation
  to real economy shocks
• in Hungary also
• Original assumptions ? and ?? produces low slope
  for marginal costs?
• Markup shocks (??)? But, usually its assumed to
  be less persistent
• Fitting macro data requires low frequency of
  price (wage) adjustments
• ? share of rule-of-thumb optimizers empirical
  results are quite robust (less than 50)
• Micro evidence for prices More frequent price
  adjustment than required for macro
• What to do then?
• understand evolution of marginal costs (mc) most
  notably wages
• Alternative solution variable capacity
  utilization (flexible inputs) at firms
• Can wage(inflation) persistence account for slow
  response of inflation?

4
Macro modelling and labour markets large focus
again

• Recent DSGE models cannot account for
  unemployment
• Can account for wage persistence (at least can
  handle)
• Other wawe of literature searching and matching
  models
• Blanchard-Gali (2006) mixing the two
  approaches
• Real labour market rigidities matter for monetary
  policy
• But Christoffel et al (2006) different wage
  bargaining matters for understanding business
  cycle facts (Nash bargain, searching) but less
  conclusive for monetary policy
• It is still a less researched area
• MNB also plans to invest in this field

5
What do we know about price and wage
persistence for Hungary

• Prices are less persistent to a monetary shock
  than wages, robust statement accross different
  models estimated for Hungary
• Source Jakab-Várpalotai-Vonnák (2006)

6
Price and wage persistence for Hungary fcast
errors of MNB

• Forecast errors if fcast is efficient
  explanatory variables should not influence fcast
  errors
• E(?X)0
• What about MNBs forecast errrors
• Jakab-Kiss-Kovács (2006) analysis
• fcast errors for CPI are unbiased
• fcast errors for wages are more likely to be
  biased and underpredicted
• THIS MIGHT SHOW AN EVEN HIGHER WAGE PERSISTENCE
  (which is still quite high)
• MNB forecasts gave right policy signals (turning
  points more or less matched)

7
Price and wage persistence for Hungary fcast
errors of MNB

• Formal test for efficacy E(?X)0
• In the longer run (4-6 quarters) Nominal wages
  lead to Positive error (higher fact than fcast)
• We still miss something! Is there even more
  persistence of prices than we estimated?
• Or wage-price connection (marginal costs
  adjustment) is not well understood
• Or Inflationary expectations are not that stable?
• Need to model labour markets in a more detailed
  way in order to arrive at better MACRO policies,
  fcasts etc
• OR is it still too early to say anything
  (Jakab-Kovács (2003) found that after an exchange
  rate shock labour market (nominal wage
  adjustment) works in the 3-5 years )

8
General modelling strategy at the MNB

• Pluralistic approach no model fits all
• Forecasting performance crucial
• Iterative method of forecasting
• model based information
• expert level information (disaggregated analyis
  also)
• Consensus view in each forecast round (quarterly)
• Focus on inflation means Labour market is key
• Lots of issues remained unresolved
• Larger human resources devoted to labour markets
  than previously
• New projects started/will start
• Job flows
• Understanding inactivity
• Modelling labour market in DSGE models

9
Labour market in the NEM model

• General features
• Nominal wage rigidities
• institutional determinants
• wage bargain (not necessarily central)
• Right-to-manage
• labour demand employers
• Employment equation from profitmax
• exogenous labour supply
• productivity and prices (LR elast1)
• Prices GDP-deflator
• Bargaining
• Nominal wage equation
• LR unemployment effect
• (elast0.14)
• level-effect on nominal wages
• LR prices and productivities (employers)
• Both equations are very slow

Labour
10
Labour market in the NEM model Some issues..

• Is unemployment measuring tightness
• Not perfectly inactivity is also cyclical and
  enter-probabilities for certain inactivity groups
  as much as for unemployed
• Problem with TREND unemployment
• employment depends on output (productivity)
• output depends on potential
• Result employment equation always predicts
  higher than actual employment growth
• capital/labour substitution is a solution
• Price variable currently GDP-deflator in NEM
• According to our recent view real unit labour
  cost deflated by CORE better in fcast
• Social security contrib. bargaining
• different than other bargaining?
• does it effect pricing?
• does it effect NAIRU?

EPRES
11
Partial labour market equations (expert info)

• VECM for employment and nominal wages in the
  private sector
• LR elasticities for wage costs price 1,
  productivity 0.6-0.7
• different models for ILO and institutional
  employment
• LR elasticities for employment value added 1,
  real wage cost -0.4
• Sectoral equations (manufacturing, services)
• Endogeneity with prices not fully solved yet
  (NEM helps)
• Issues How to treat minimum wage effects?
• How to treat bonuses, and how to disentangle
  bonus shocks from minimum wage shocks?
• Do bonuses translate into income and consumption?

12
The way ahead new estimated DSGE model

• MNB is now building a new DSGE model
• Key features
• Two sectors traded and nontradeds
• Price and wage stickiness álá Calvo
• indexation álá Smets-Wouters
• habit in consumption
• export market share rigidities
• costly capital, labour and imported intermediate
  adjustment
• General monetary policy rule incorporating both
  the crawling peg and inflation targeting
• Modified UIP álá Schmitt-Grohe and Uribe
• Currently calibrated version, estimation underway
• Estimation method bayesian estimation with
  structural break
• Labour market quite standard yet
• In the future different labour market setup
  (matching and/or effective wages)
• Christoffel et al (2006) for Germany, different
  labour market setups help in better understanding
  business cycle features of these models, but for
  monetary policy design its not that important
• Other view Smets-Wouters (2006) less sceptical

13
Challenges to our models K/L substitution

• As mentioned before wages might be even more
  persistent, what are the reasons?
• Other question Capital/Labour substitution might
  seem to be underway
• Problem no simple production fn can capture it
• A good measure K/(LA) against relative prices
  (Comp/P/Price of capital)
• price of capital proxied by price of investment
• Is this due to economic policy (e.g. increased
  public wages, minimum wages etc.)
• Or is it a natural story
• 95-97 FDI flows due to
• cheap labour and lower social
• securities
• 2000 sectoral story
• mosty textiles

14
Challenges to our models K/L substitution

• As mentioned before wages might be even more
  persistent, what are the reasons?
• Other question Capital/Labour substitution might
  seem to be underway
• Problem no simple production fn can capture it
• A good measure K/(LA) against relative prices
  (Comp/P/Price of capital)
• price of capital proxied by price of investment
• Is this due to economic policy (e.g. increased
  public wages, minimum wages etc.)
• Or is it a natural story
• 95-97 FDI flows due to
• cheap labour and lower social
• securities
• 2000 sectoral story
• mosty textiles

15
Challenges to our models Effective labour

• In our models ILO unemployment measures labour
  market tightness
• Is it a good measure?
• Cseres-Gergely and Chassambouli (2006) found
  large flows between inactives and employed
• Demographic factors low activity rate might be
  due to demography and initial conditions
  (transition) and early retirement policies
• Lamo (2006) modelled new and old industries
  for Poland and Estonia
• Transition is low first high U, then sharp
  decline
• This might also happened in Hungary during 90s
• Are we now in normal times when U is a measure
  for matching probabilities?
• Maybe not Sectoral reallocation of labour is
  still an issue

16
Challenges to our models Substitution between
public and private labour

• In our models we treat them as perfect
  substitutes in labour market tightness
• (U in the private wage equation)
• public wages are exogenous
• Public wage hike has only second round effects to
  private wages (though U, demand etc..)
• Is it a good description?

17
Thank you
18
Labour Market Phillips Curve and main mechanisms
GDP on employment
Wage cost on employment
Productivity and prices on Wages
Unemployment on Wages
19
Need for add-factoring employmentUnless high
employment growth