The Impact of Macroeconomic Fluctuations on Training Decisions - - PowerPoint PPT Presentation

Introduction Data Estimation Methodology Results Model Policy More Charts

The Impact of Macroeconomic Fluctuations

• n Training Decisions

Vincenzo Caponi Ryerson University, IZA, RCEA Cevat Burc Kayahan Acadia University Miana Plesca University of Guelph

CLSRN − HRSDC Workshop 18-19 November 2008, Toronto

Introduction Data Estimation Methodology Results Model Policy More Charts

Motivation

• Training should take place when output is low.

− smaller opportunity costs; deJong and Ingram (RED 2001) Dellas and Sakellaris (1996) Devereux (JOLE 2000).

• Training should take place when output is high.

− adoption of new technologies which may require training to operate; − financing easier, outside options better; King and Sweetman (RED 2002).

Introduction Data Estimation Methodology Results Model Policy More Charts

Our contribution

• We provide a unifying framework where both channels are

identified

• Investigate how training decisions are affected by
• Aggregate macroeconomic fluctuations
• Relative sectoral output fluctuations
• Idea: the persistence of idiosyncratic sectoral shocks

should be higher than that of the aggregate shocks; this links with the training decision.

Introduction Data Estimation Methodology Results Model Policy More Charts

Main findings

• 1. Aggregate macroeconomic fluctuations have a negative

impact on firms’ propensity to train.

• 2. Relative sectoral fluctuations have a positive impact on

firms’ propensity to train. We illustrate these two channels in a search model with random matching, human capital acquisition, and sectoral and aggregate shocks. Policy implications.

Introduction Data Estimation Methodology Results Model Policy More Charts

Data: Aggregate statistics and WES

Aggregate output statistics (StatsCan) 1980-2007 HP filtered Real 2000 dollars WES 1999-2005 Unit of analysis: the firm WES follows NAICS with small differences in aggregation Timing of WES: March 31st to April 1st Definition of training: formal classroom training

Introduction Data Estimation Methodology Results Model Policy More Charts

GDP , HP-filtered GDP and HP trend

GDP

6.000e+11 8.000e+11 1.000e+12 1.200e+12 gdp 1980 1990 2000 2010 year

• 2

. e + 1 0- 1 .0 e + 1 1 .0 e + 1 0 2 .0 e + 1 H P _ g d p _ 1 1980 1990 2000 2010 year 6 . e + 1 1 8 . e + 1 1 1 . e + 1 2 1 . 2 e + 1 2 H P _ g d p _ s m _ 1 1980 1990 2000 2010 year

Introduction Data Estimation Methodology Results Model Policy More Charts

Table: Sectors Considered in the Analysis

Sector Size (%) Forestry and Mining 0.05 Construction 0.13 Transportation, Warehouse, Wholesale Trade 0.13 Information, Communication and Utilities 0.10 Finance and Insurance 0.07 Real Estate 0.06 Business Services 0.10 Education and Health 0.04 Manufacturing 0.21 Retail Trade and Consumer Services 0.11

• N. Obs.

5535

Introduction Data Estimation Methodology Results Model Policy More Charts

Charts

Sector to GDP ratio

0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 1999 2000 2001 2002 2003 2004 2005

Foresty & Mining Construction Transportation, Warehouse, Wholesale Trade Information, Communication and Utilities Finance and Insurance Real Estate Bussiness Services Education & Health Manufacturing Retail Trade and Consumer Services

Introduction Data Estimation Methodology Results Model Policy More Charts Training incidence

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1999 2000 2001 2002 2003 2004 2005

Foresty & Mining Construction Transportation, Warehouse, Wholesale Trade Information, Communication and Utilities Finance and Insurance Real Estate Bussiness Services Education & Health Manufacturing Retail Trade and Consumer Services Average

Introduction Data Estimation Methodology Results Model Policy More Charts Sector to GDP ratios (2b)

0.04 0.045 0.05 0.055 0.06 0.065 Construction 17.68% 18.26% 18.87% 17.54% 17.37% 17.03%

Foresty & Mining Transportation, Warehouse, Wholesale Trade Bussiness Services Manufacturing

Introduction Data Estimation Methodology Results Model Policy More Charts Training incidence (2b)

0.2 0.25 0.3 0.35 0.4 0.45 0.5 1 2 3 4 5 6 7

Transportation, Warehouse, Wholesale Trade Bussiness Services Manufacturing Foresty & Mining

Introduction Data Estimation Methodology Results Model Policy More Charts

Estimation framework

Dit binary training indicator Pit the probability of firm i to train workers in period t Pit|ωt = Pr(Dit = 1|ωt) = E[Di|ωt] ωt a collection of Zt firm characteristics Xt sector or overall economy characteristics. We model the conditional expectation using the logistic distribution E[Di|ωt] = Λ(αi +βZit +δXit +uit)

[Further Evidence]

Introduction Data Estimation Methodology Results Model Policy More Charts

Table: Sample Statistics

Variable Description Mean Std Dev Classroom Training Indicator 0.340 0.474 % Workforce Trained 0.213 0.446 Overall Training Indicator 0.577 0.494 Firm size Number of workers employed by the workplace 16.7 49.7 Innovation Adoption of innovation and/or new technology by the workplace 0.489 0.499 Unionized Indicator whether the workplace is unionized 0.057 0.232 Multiple loc. Indicator whether the workplace belongs to a multiple-location firm 0.455 0.498 Market The most dominant sales market of the firm Local 0.855 0.351 Canada 0.095 0.292 World 0.049 0.217 Skill % of workforce in skill groups Administrative 0.197 0.283 Managers 0.202 0.231 Others 0.074 0.225 Professionals 0.059 0.169 Sales 0.122 0.249 Technicians 0.148 0.263 Production 0.198 0.312

Introduction Data Estimation Methodology Results Model Policy More Charts

Main results

Table: The Impact of Aggregate Fluctuations and Sectoral Deviations

• n the Incidence of Training

Variables Coefficients P-value Innovation 0.584 Market Canada 0.086 Market World 0.472 Firm size 0.559 Multiple locations 0.156 Unionized

• 0.12

% Administrative 0.552 % Managerial 0.573 % Other 1.145 % Sales 0.283 % Production 0.637 % Technical 0.13 GDP deviations

• 0.008

Sector to GDP ratio 0.03 Number of observations 8881

Introduction Data Estimation Methodology Results Model Policy More Charts

Table: Impact of Fluctuations on Training Incidence: Controlling for Previous Training Benefits

Variables Coefficients P-value Innovation 0.721 0.000 Market Canada 0.361 0.000 Market World 0.502 0.000 Firm size 0.602 0.000 Multiple locations 0.094 0.000 Unionized

• 0.099

0.000 %Administrative 1.669 0.000 %Managerial 1.395 0.000 %Other 1.616 0.000 %Sales 0.819 0.000 %Production 1.521 0.000 %Technical 0.914 0.000 GDP deviations

• 0.009

0.000 Sector to GDP ratio 0.049 0.000 MBt−1 0.104 0.001 MBt−2 0.277 0.000

Introduction Data Estimation Methodology Results Model Policy More Charts

Table: Impact of Fluctuations on Training Intensity: Continuous Training Measure (% workforce trained)

Variables Coefficients P-value Innovation 0.604 0.000 Market Canada 0.080 0.000 Market World 0.457 0.000 Firm size 0.536 0.000 Multiple locations 0.094 0.000 Unionized

• 0.118

0.000 % Administrative 0.552 0.000 % Managerial 0.535 0.000 % Other 1.127 0.000 % Sales 0.268 0.000 % Production 0.630 0.000 % Technical 0.112 0.000 GDP deviations

• 0.006

0.000 Sector to GDP ratio 0.001 0.000

Introduction Data Estimation Methodology Results Model Policy More Charts

Table: Impact of Fluctuations on Training Incidence: Adding OJT to CT in the Definition of Training

Variables Coefficients P-value Innovation 0.49 0.00 Market Canada 0.328 0.00 Market World

• 0.356

0.00 Firm size 0.352 0.00 Multiple locations

• 0.077

0.00 Unionized

• 0.307

0.00 % Administrative 0.209 0.00 % Managerial 0.051 0.00 % Other

• 0.037

0.00 % Sales 0.464 0.00 % Production 0.337 0.00 % Technical

• 0.048

0.01 GDP deviations

• 0.0001

0.59 Sector to GDP ratio 0.061 0.00 Number of observations 6427

Introduction Data Estimation Methodology Results Model Policy More Charts

Simple Mortensen-Pissarides model with sectoral and aggregate shocks and training

• Random meeting between vacancies v and searchers u

governed by the meeting function m(v,u).

• One firm is one sector.
• Once there is a meeting between a worker and a firm the

match-specific productivity x is realized. (Note: x is the sector-specific idiosyncratic shock).

• The training decision takes place, simultaneous with the

decision whether to form a productive match.

• Training increases human capital according to the human

capital function exp(H(x)) at a cost c(X).

• If no training is offered, H(x) = 0.
• If α(x) = x ·exp(H(x)) is above the reservation

productivity value, a match is created.

Introduction Data Estimation Methodology Results Model Policy More Charts

• After a match is created, the match can be hit by an

aggregate productivity shock p that arrives at rate λ.

• The productivity shock changes the worker specific

productivity α(x). If the productivity decreases below the ex-post reservation level, the match is dissolved.

• Productive matches can be destroyed exogenously at rate

δ.

• Wage are set by Nash bargaining.

Introduction Data Estimation Methodology Results Model Policy More Charts

Hazards

mi(v,u) = m(1, u v )v ≡ q(θ)v where θ = v u .

• Rate at which vacancies are filled:

qf = q(θ)

b

r dF(α),

(1)

• Rate at which unemployed workers find a job

qw = q(θ)θ

b

r dF(α).

(2)

Introduction Data Estimation Methodology Results Model Policy More Charts

Value functions

• Value of a match to an employer

rJ(α) = pα −w(α)+λ

b

r [J(x)−J(α)]dF(x)−λF(r)J(α)

(3)

• Value of a match to the worker

rW(α) = w(α)+λF(r)[U −W(α))] (4)

• The value of setting a vacancy

rV = −pc +qf[Je −V]. (5)

• The value of being unemployed

rU = b +qw[W e −U], (6)

Introduction Data Estimation Methodology Results Model Policy More Charts

Policy Implications

From a policy point of view, it is important to understand that documented trends in training incidence may be a result of

• ptimal responses to macroeconomic circumstances.

Introduction Data Estimation Methodology Results Model Policy More Charts

More charts

% Workforce trained

0.1 0.2 0.3 0.4 0.5 0.6 1999 2000 2001 2002 2003 2004 2005

Foresty & Mining Construction Transportation, Warehouse, Wholesale Trade Information, Communication and Utilities Finance and Insurance Real Estate Bussiness Services Education & Health Manufacturing Retail Trade and Consumer Services

Introduction Data Estimation Methodology Results Model Policy More Charts Sector to GDP ratios (2a)

0.07 0.072 0.074 0.076 0.078 0.08 0.082 1999 2000 2001 2002 2003 2004 2005

Finance and Insurance Real Estate

Introduction Data Estimation Methodology Results Model Policy More Charts Training incidence (2a)

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1 2 3 4 5 6 7 Finance and Insurance Real Estate

Introduction Data Estimation Methodology Results Model Policy More Charts Sector to GDP ratios (2c)

0.1 0.105 0.11 0.115 0.12 0.125 0.13 0.135 1999 2000 2001 2002 2003 2004 2005 Information, Communication and Utilities Education & Health Retail Trade and Consumer Services

Introduction Data Estimation Methodology Results Model Policy More Charts Training incidence (2c)

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1 2 3 4 5 6 7 Information, Communication and Utilities Education & Health Retail Trade and Consumer Services