WSF Working Paper Series

GlobLabWS

#2/2018

November 2018

De-Globalisation, Welfare State Reforms and Labour Market Outcomes

Hassan Molana, Catia Montagna,

George E. Onwordi

FundingERA-Net Plus Funding

Grant Agreement Number 618106

WSF Working Paper SeriesNORFACE Welfare State Futures

Scientific Coordination OfficeZiegelstr. 13c, 10117 Berlin

+49 (0)30 2093 1456info.wsf.sowi (a) hu-berlin.de

welfarestatefutures.org/working-papers

http://welfarestatefutures.org/publications

De-Globalisation, Welfare State Reforms and Labour Market Outcomes

Hassan Molana University of Dundee

Catia Montagna University of Aberdeen

George E. Onwordi University of Aberdeen

October 2018

Abstract: De-globalisation, by raising trade barriers in an economy characterised by vertical linkages in production and imperfect goods and labour markets, is shown to increase unemployment across skill levels, deteriorate job-skill-match quality, and reduce aggregate income. These effects are enhanced when the firm-level productivity distribution is more skewed towards less productive firms, and are not necessarily moderated by maintaining frictionless mobility of capital across borders. A Flexicurity reform of a liberal welfare state is shown to dampen the adverse effects of de-globalisation. The effectiveness of the reform is strengthened if the package includes a rise in public investment in job-search/matching efficiency.

Keywords: Flexicurity; Globalisation; Skill Mismatch; Unemployment; Welfare State JEL Codes: F16, F6 Corresponding Author: Catia Montagna, Department of Economics, Aberdeen Business

School, Edward Wright Building, Old Aberdeen, AB243QY; c.montagna@abdn.ac.uk.

Acknowledgements: We would like to thank Stephen Boyd, Nils Braakmann, Holger

Görg, Sebastian Könings, Steve Matsus, Fredrik Sjöholm, John Skåtun, Allan Sørensen and participants at the NORFACE Welfare State Futures Programme Final Conference at the EUI in Fiesole and at the GlobLabWS final workshop in Aberdeen for useful comments and suggestions. This work was supported by the NORFACE ERA-NET (New Opportunities for Research Funding Agency Co-operation in Europe Network) Welfare State Futures Programme, Grant Number 462-14-120. The usual disclaimer applies.

mailto:c.montagna@abdn.ac.uk

1

1. Introduction The perception that international trade has been a source of economic dislocations with adverse effects on labour market outcomes, particularly of low skilled workers in industrial economies, is arguably one of the causes of the backlash against globalisation. According to many analysts, this phenomenon underpins the emergence of protectionist stances and political developments such as Brexit. This paper aims to examine the effects of increases in trade costs on labour market outcomes and how they are affected by labour market reforms. We argue that raising trade barriers (now often referred to as de-globalisation) may in fact not only have a negative impact on the level of unemployment, but also on the nature of employment, as reflected by the existence of job-skill mismatch – particularly if higher trade costs affect a country’s participation in the global vertical production chain with negative consequences for aggregate productivity, a theme that has been central to debates about the potential implications of Brexit for the UK – and that welfare state and labour market policies may play a key role in determining these effects. Despite the employment gains achieved since the Great Recession, the persistence of skill mismatch, increasingly in the form of over-education, is a central challenge facing European labour markets. Albeit with differences between workers’ characteristics (age, gender) and across countries, evidence suggests that over-education affects between 20% and 50% of the employed population in Europe (ILO, 2014). The existence of gaps between the skills possessed by individuals and those required by jobs has important potential consequences. Skill underutilisation, in particular, is likely to reduce an economy’s productivity and, arguably, it might have been a contributing factor to the weak recovery experienced in many European countries (ECB, 2012).1 The increase in the incidence of mismatch has been ascribed to the combined effects of technological progress and changing trade patterns on the demand side and increasing levels of educational attainments on the supply side.2 Differences across countries, in turn, have been attributed to factors such as the heterogeneity across labour market institutions (Brunello et al., 2007; Adalet McGowan and Andrews, 2015b), differences in the extent of public expenditure on active labour market policies (European Commission, 2013; Marsden et al., 2002), and productivity differentials (Adalet McGowan and Andrews, 2015a). To address these issues, we construct a general equilibrium model of an open economy characterised by vertical linkages in production, search frictions and two-sided heterogeneity in the labour market. Firms exhibit different productivity levels whilst workers are endowed 1 Although, as explained by some (e.g. Green et al., 1999), mismatch might reflect temporary imbalances in the labour market that are likely to fizzle out without the need for any specific policies, the recent evidence of its high and persistent nature may suggest otherwise. This is reflected in the fact that a key focus of the EU 2020 project – and the recent emphasis by Draghi (2014) – is the deployment of policies that facilitate the matching workers’ skills with those required by their employers. 2 In a recent study, Adalet McGowan and Andrews (2015b) find that the availability of skilled workers does not necessarily result in a lower mismatch, indicating that policies that induce higher educational attainment may not inevitably translate into better assortative matching between firms and workers.

2

with different skill levels – as in, e.g. Albretch and Vroman, (2002) partial equilibrium framework. Firms use labour and capital to produce varieties of an intermediate good sold both domestically and internationally. A central assumption of the model is that whilst high-tech jobs can only be performed by skilled workers, low-tech jobs can be performed by both skilled and unskilled workers. This gives rise to skill mismatch and, to capture its potentially transitory nature, we allow for job-to-job transition whereby mismatched workers search on-the-job for high-tech jobs whilst employed in low-tech ones. A distinct contribution of the paper is that, unlike previous studies, our framework introduces adjustments in the labour force by endogenising participation – an important feature that has so far been neglected in models that examine the labour market effects of economic integration. There are three main reasons why addressing this omission is important. First, as documented by, e.g., Elsby et al. (2015), roughly one-third of the variations in unemployment over time can be accounted for by movement in and out of the labour force, and this plays a key role in driving aggregate labour market outcomes. Second, recent empirical evidence, e.g., Gaddis and Pieters (2012), Autor et al. (2013) and Cooray et al. (2017), shows that economic integration has had a significant effect on labour force participation decisions. Finally, as we demonstrate in this paper, these adjustments are crucial both when assessing the impact of shocks (such as raising trade costs) and from a policy perspective.3 We consider a broad menu of policy instruments and capture the complex interaction between their effects when they are combined in a policy reform package. From a methodological standpoint, since changes in individual policies may have opposite effects on the equilibrium values of the variables, the extent to which they are altered relative to each other is an important determinant of the net impact of a given reform package. To this end, we start by calibrating the baseline steady-state solution of the model to reflect the main characteristics of the UK economy in recent years, which can be thought of as representing the long-run equilibrium of a liberal welfare state regime. We then compare this solution with that obtained by implementing a reform package that targets unemployment benefit, firing and vacancy creation costs in the direction of and by a proportion consistent with their corresponding Danish counterparts (a typical example of a flexicurity system). Notably, such a reform package would entail reducing the degree of flexibility of the labour market and increasing expenditure on both active and passive labour market policies. We then expand the reform package to include increases in the level of expenditure on employment services – a policy that has received growing attention as a cost-effective means to reduce labour market

3 In models with full and exogenous participation (see e.g., Cacciatore et al, 2016; Cacciatore, 2014; Felbermayr et al., 2011a; Helpman and Itskhoki, 2010), changes in unemployment following a shock occur primarily through movements in vacancy creation activities that shape market tightness and workers’ job finding probabilities. Making participation endogenous, e.g. by allowing the household to decide on participation level of its members, renders unemployment sensitive to fluctuations in both labour force and vacancies.

3

frictions4 but which has arguably become less prominent in the UK in recent years (Davies, 2018). Our model shows that an increase in trade frictions, as might arise from a ‘no-deal Brexit’, can have adverse consequences for the level of economic activity, unemployment and job match quality that arise from the negative effects of higher trade costs on firm profitability and labour demand. Crucially, we find that the endogeneity of labour market participation matters greatly in determining the effects of policy. For instance, increases in unemployment benefits (a typical passive labour market policy) can perform as an activation policy and have expansionary effects, contrary to conventional views that portray it as a distortionary policy that harms employment via higher labour costs. More generally, our model suggests that despite involving stronger restrictions on firing, a reform package that involves higher expenditure on active labour market policies can increase the level of economic activity and welfare via aggregate supply and aggregate demand effects that stimulate both labour market participation and job creation and result in an improvement in the quality of job matching (by increasing transition rates both from unemployment and from job-to-job) and in a fall in unemployment across the skill spectrum. Such a reform is shown to play a crucial role in mitigating the potentially adverse effects of negative shocks such as Brexit on the UK economy. Finally, we carry out counterfactual experiments to examine the effects of changes in the productivity distribution of firms or in the degree of capital mobility frictions. Our model predicts that the negative impact of a higher trade cost is greater when the density of firms at high productivity levels is lower. We also demonstrate the importance of the interaction between the degree of trade openness and capital mobility frictions. Higher capital mobility frictions, by increasing the cost of capital, triggers a substitution away from capital and toward labour in production and hence in higher wages and employment; as a result, the negative impact of raising trade costs are not necessarily moderated by maintaining frictionless mobility of capital across borders. The extant literature on the effects of international economic integration on the labour market is vast and varied. A strand of this literature, to which this study is closely related, focuses on the productivity and unemployment effects of economic integration, but does not reach a clear consensus. For instance, Felbermayr et al. (2011a) and Cacciatore (2014), amongst others, show that higher trade integration reduces unemployment by inducing a reallocation of resources towards more productive firms. By contrast, Helpman and Itskhoki (2010) and Helpman et al. (2010) find that trade openness can potentially result in higher unemployment despite leading to higher firms’ profitability. Moore and Ranjan (2005) argue that trade liberalisation can lead to a higher unemployment rate of the unskilled, whereas its

4 See, e.g. Lehman and Kluve, 2010. The OECD (2015) has endorsed polices that improve matching processes as the most cost-effective form of ALMP consistent with the evidence documented in Riley et al. (2011).

4

effect on aggregate unemployment is ambiguous. A major advancement of our paper in relation to these studies is to allow for the emergence of labour market mismatch with endogenous participation. The effects of openness on the quality of labour market matching has received relatively little attention in the literature. At an empirical level, Davidson et al. (2014) and Krishna et al. (2014) find evidence of improved match quality as a result of globalisation. At a theoretical level, building on the partial equilibrium framework with two-sided heterogeneity developed by Albrecht and Vroman (2002), Davidson et al. (2008) study the effects of trade liberalisation but focus on firms’ export decisions. A similar approach is found in Arseneau and Epstein (2017) who study the effects of openness on labour market outcomes and argue that mismatched employment helps to moderate the higher aggregate unemployment consequences of offshoring. In addition to allowing for the job-search decisions of the unemployed to reside with the household, our paper differs from these studies in that it considers the transitory nature of mismatch by allowing for job-to-job flows that we show to play a key role in driving match quality. Furthermore, whilst these papers consider two types of firms, our assumption of a continuous productivity distribution allows us to consider the interaction between the shape of the distribution and average industry productivity, policy, and changes in trade costs. Moore and Ranjan (2005) also study the impact of globalisation on the unemployment outcomes of workers with different skills but focus on a labour market in which only perfect job matches exist in equilibrium. Finally, by considering the interaction between labour market policies and institutions and the degree of international trade openness, our work is also closely related to, e.g., Helpman and Itskhoki (2010) and Coşar et al. (2016) but is distinguished from them by its use of an explicit definition of workers’ heterogeneity and by providing new explanations for how (de-) globalisation interacts with a multiplicity of labour market policies to drive mismatch and unemployment of different categories of workers. The rest of the paper is organised as follows. Section 2 sets up the model and Section 3 describes the model calibration. The effects of de-globalisation and reforms are explained in Sections 4 and 5 respectively. Section 6 compares the pre- and post-reform effects of de-globalisation, where we also examine the roles of firms’ heterogeneity, public investment in improving job-matching efficiency, and capital mobility frictions. Section 7 concludes the paper. 2. The Model We construct a dynamic model of a small open economy in which a representative household’s members supply skilled and unskilled labour. Capital serves the dual purpose of wealth accumulation and factor of production and is allowed to be internationally mobile. In an upstream sector, monopolistically competitive firms with firm-specific productivities use capital and skilled and unskilled labour to produce varieties of an intermediate input which

5

they export as well as sell domestically to a downstream sector. The latter combines the domestic and imported varieties to produce a homogenous final good under competitive conditions. The labour market clears through a search and matching process. The government implements labour market policies and uses a lump-sum tax levied on the household to balance its budget. 2.1. Households There is a representative household with a continuum of infinitely-lived members whose measure is normalised to unity. Household members are either skilled or unskilled with their respective mass treated as exogenous and denoted by Z and 1 − Z. At any point in time t, each type is assumed to be economically active (participating in the labour force) or inactive. Denoting the latter states by X and L, respectively, and using the superscripts s and u to refer

to skilled and unskilled workers, it follows that s st tX L Z+ = and 1u ut tX L Z+ = − . There are

two types of tasks, low-tech and high-tech. Those participating in the labour force are either employed, or unemployed and searching for a job and are denoted by N and S, respectively.

The unskilled can only search for and be employed in low-tech task jobs, hence ,u ul ult t tX S N= +

where the superscripts ul refers to unskilled in low-tech task jobs. The skilled can search for and be employed in either task. Hence, respectively denoting by superscripts sl and sh those

who go for the low- and high-tech task jobs, s sl sht t tX X X= + . While sltX are assumed to have

opted for low-tech task jobs in order to exit from the unemployment pool, they will engage in on the job search (OTJS) for the high-tech task jobs.5 Therefore, we also use sl sl slt t tX N S= +

and sh sh sht t tX N S= + to partition participation of skilled workers into employed and searching.

All newly-formed job matches at any time t are assumed to start working at the beginning of the following period. Thus, as far as the household is concerned, the three employment types evolve as follows6

( )1 1ul l ul l ult t t tN N q Sη+ = − + , (1) ( ) ( )1 1 1sh h sh h sh l h slt t t t t tN N q S e q Nη η+ = − + + − , (2) ( )( )1 1 1sl h l sl l slt t t t tN eq N q Sη+ = − − + , (3) where jη and jq are, respectively, the exogenous job destruction (or match separation) rate and the endogenous probability of a job match (job-finding rate), the superscript j = h, l refers

to low- and high-tech task jobs, and [ ]0,1e∈ is the efficiency measure of OTJS. Denoting the

high-tech and low-tech matched jobs respectively by ht and lt , it follows that

5 This is a form of competitive search similar to Arseneau and Epstein (2014). 6 For simplicity, we abstract from the intensive margin of employment.

6

( )( )1h h sh l slt t t tq S e Nη= + − and ( )l l ul slt t t tq S S= + . Equation (1) shows that the mass of unskilled workers who are employed at the beginning of time t +1 consists of those who

survived their ‘match separation’, i.e. ( )1 ,l ultNη− and the new matches l ult tq S . Equation (2) states that the mass of skilled workers employed in high-tech jobs consists of those who

survived their match separation in high-tech jobs, ( )1 h shtNη− , the new job matches through direct search, ,h sht tq S and the previously mismatched workers who found high-tech jobs through

OTJS ( )1 l h slt te q Nη− . Equation (3) shows that mismatched employees consist of those who are unable to find high-tech job matches through OTJS with probability ( )1 hteq− but have survived their match separation in their existing mismatched low-tech job with a probability

( )1 lη− , and the new mismatched matches l slt tq S . The household pools income from all sources and faces the budget constraint,7

( )*ul ul sl sl sh sh D Dt t t t t t t t t t t t tt tt tbC I T w N w N w N S r K r K K+ + = ++ Π+ + + + − , (4) where: C is consumption; I is gross investment; T is the lump-sum tax paid to the government;

, , , ,jw j ul sl sh= is the negotiated wage rates paid respectively to unskilled workers in low-tech job, skilled workers in mismatched low-tech jobs, and skilled workers in high-tech jobs; b is

the unemployment benefit paid to those who are actively searching for jobs, ;ul sh slS S S S= + + Π is firms’ profits which is distributed to households (to be clarified later); K is the capital stock held by the household sector; DK is firms’ demand for capital stock; and r and *r are respectively the domestic and foreign rate of return on capital. The budget constraint above reflects the economy’s international borrowing/lending of capital with an inflow (outflow) of

corresponding to ( )0 0 .Dt tK K− > < The stock of capital depreciates at the constant rate δ leading to the capital accumulation process

( )1 1t t tK I Kδ+ = + − . (5) The instantaneous utility function of the household is assumed to be

( ) ( ) ( ) ( )u u sl sl sh sht t t t tU U C A X A X A X= − − − , (6) where ( )tU C is the utility from consumption and ( )j jtA X represents the disutility of participation (not enjoying leisure) of the relevant worker type. Treating the paths for

{ }*, , , , , , | 0, ,j l h Dt t t t t t t t tw b r r q q K T tΠ ≥ and the initial condition { }0 0 0 0, , ,ul sl shK N N N as given, the household chooses the optimal paths for { }1 0, , |tt t jC K X t+ ≥ to maximise the expected value

7 Following Merz (1995), we assume full risk sharing within the household. While this is of course a strong assumption, it enables us to circumvent the complications that could result from the heterogeneity of household members and their various employment states.

7

of 0

tt

tUβ

∞−

=∑ . The first order conditions for the intertemporal maximisation problem can be

shown to imply the standard Euler equation governing the path of consumption

( ) ( )( )1 11t t t tU C E U C rβ δ+ +′ ′= + − , (7)

where ( )0,1β ∈ is the subjective time preference discount factor, and the following relationships which determine the relative marginal disutility of work governing the household’s labour market participation decisions

( )( )

( )( )

( ) ( )( ) 11

1 11 1

1 1

1u u u u u ut tl

lt t t tl

tult

ttt

t t

b q EA X A X A X

wU C U C U C

bq

η+

+

+ ++ +

+ +

′ ′ ′ = Λ + −

′ ′ ′

− − −

, (8)

( )( )

( )( )

( )( ) ( )( ) ( )

( )( )

11 1

1

1 11 1

1 1 1

11 1 1 ,

lt t t

h lt l

t tl

sl sl sl slt tsl

t tt t

sl sl sh

t

sht t

t t

A X A Xw

U C U C

A X A XU C U C

e

b q E

eqb b

qη

η

++ +

+

+ +

+ +

++ +

+

′ ′− Λ

′ ′

′ ′ − + −

′ ′

−

= −

− + −

(9)

( )( )

( )( )

( ) ( )( )

1 11 1

11

1 1

1 hh

t t t tht

sh sh sh sh sh sht t tsh

t tt t t

A X A X A Xw

U C U C U Cb q E b

qη

++

+ ++ +

+ +

− ′ ′ ′ Λ + −

′ ′ ′

− = −

. (10)

where we have used (7) to define ( ) ( )1 1t t tU C U Cβ+ +′ ′Λ = as the stochastic discount factor. Each equation expresses the net marginal cost of the relevant members’ participation in terms of their expected net marginal benefit of securing a lasting job match. 2.2. Vacancies and matching We assume that two types of ‘specialised’ hiring agencies, labelled low- and high-tech, act as intermediaries between workers and firms operating in the intermediate good sector to meet their low-skill and high-skill labour demand. The low-tech agencies can engage the unskilled as well as the skilled workers willing to opt for a low-tech job whereas the high-tech agencies only consider skilled workers. The presence of OTJS implies that skilled workers who are employed through the low-tech agency can search while on the job for a high-tech job. These

agencies create low-tech and high-tech job vacancies , , ,jV j h l= which we assume to have a

unit cost of ,jc and fill them following the process governing the search and matching frictions. Recall that the existing low-tech and high-tech job matches are subject to

exogenously determined separation (job destruction) rates jη . This amounts to a fixed firing

rate and it is assumed that a fixed firing cost per worker of f is incurred by the agencies. Below we describe the bargaining process between each type of agency and worker that determines the respective wages.

8

2.2.1. Low-tech job agency At any time t, the aggregate number of matches in the low-tech segment of the labour market

is determined by the matching function ( ),l l ul sl lt t t tm S S V= + which is assumed to satisfy the standard properties described by Pissarides (2000). We define market tightness and the

probability of filling a low-tech vacancy (hiring rate) respectively by ( )l l ul slt t t tV S Sθ ≡ + and .l l lt t tVρ ≡ It is also useful to define ( )sl sl ul slt t t tS S Sξ ≡ + , the fraction of low-tech job

searchers who are skilled, so that the effective probability that a low-tech agency matches with

a skilled job searcher is sl sl lt t tρ ξ ρ≡ , while ( )1ul sl lt t tρ ξ ρ≡ − is the effective matching probability it has with an unskilled job-searching worker. Thus, from the agency’s perspective, employment of unskilled and skilled worker evolves according to

( )1 1ul l ul ul lt t t tN N Vη ρ+ = − + , (11) ( )( )1 1 1sl h l sl sl lt t t t tN eq N Vη ρ+ = − − + . (12) Denoting by lH the effective man-hour obtained from the ultN and

sltN workers, the

agency’s revenue from these workers is l lt tw H , where lw is the wage rate it receives per worker

from the firms that hire these workers’ services. Thus, the agency’s temporal profit is

( ) ( ), ,l sl l l l ul ul sl sl l l l ul slt t t t t t t t t t tu tlt V w H w N w N c V NN N f Nπ η = − + + + + , (13) where the term in square brackets on the right-hand-side consists of the costs from employment, vacancy creation, and firing. Denoting the job value for the agency at each period by

( ), lult stN N , it follows that the solution to the maximisation of its present value solves the Bellman equation

( ) ( ) ( )1 11max ,, , ,lt

ul usl l sl l slt t t

l ult t t

Vtt tVN N N N NENπ + + + = + Λ . (14)

Let ( ),ul ul sl ult t t tN N N≡ ∂ ∂ and ( ),sl ul sl slt t t tN N N≡ ∂ ∂ . The marginal condition that removes any incentives for other competing agencies to be set up is

1 1 1l ul ul sl slt t t t t t tc E ρ ρ+ + + = Λ + , (15)

which eliminates profits from vacancy creation by equating its unit cost with the expected present value of its marginal benefit, given by the weighted average of marginal gains from employing unskilled and skilled workers. The latter evolve according to the partial derivatives

of equation (14) with respect to ulN and slN ,

( ) 1 11l

ul l ul l l ultt t t t t tul

t

Hw w f EN

η η + +∂

= − − + − Λ∂

, (16)

9

( )( ) 1 11 1l

sl l sl l h l sltt t t t t t tsl

t

Hw w f eq EN

η η + +∂

= − − + − − Λ∂

, (17)

which state that the marginal gain (or the surplus) of a job match to the agency is given by the marginal revenue of a worker net of the wage rate it receives from the agency plus the expected discounted continuation value of the job. 2.2.2. High-tech job agency From the agency’s perspective, its employment of skilled workers evolves as

( )1 1sh h sh h ht t t tN N Vη ρ+ = − + , (18) where h h ht t tVρ ≡ is the vacancy filling probability and

htV is the number of vacancies the

agency creates. The aggregate number of matches is determined by the matching function

( )( )1 ,h h sh l sl ht t t tm S e N Vη= + − where ( )1sh l slt tS e Nη+ − is the number of high-tech job seekers consisting of those who search directly and those who engage in OTJS while employed

in a low-tech job. Thus, market tightness is ( )( )1h h sh l slt t t tV S e Nθ η≡ + − . Similar to the low-tech agency case, the temporal profit of the agency is

( ),h sh h h h sh sh h h h sht t t t t t t t tV w H w N c VN f Nπ η = − + + , (19)

where hH is the effective man-hour supplied by shtN workers and hw is the wage rate the

agency receives for a worker from firms that employ these workers’ services. The maximized

job value ( )shtN should then solve the Bellman equation ( ) ( ) ( )1 1max ,h

t

h h sht t t

sh h

vt ts

t V E NN Nπ + + = + Λ , (20)

and the marginal condition that eliminates profits from vacancy creation is 1 1

h h sht t t t tc Eρ + += Λ , (21)

where ( )sh sh sht t tN N≡ ∂ ∂ whose evolution is given by the derivative of equation (20) with respect to shN . Hence,

( ) 1 11h

sh h sh h h shtt t t t t tsh

t

Hw w f EN

η η + +∂

= − − + − Λ∂

. (22)

2.3. Wage determination For simplicity, we use the conventional Nash bargaining approach and assume that wages are negotiated so as to maximise the weighted product of each party’s match surplus. Given that profits from vacancy creation are eliminated, the match surpluses for the agencies are

, , , .jt j ul sl sh= The corresponding surpluses for each type of worker, denoted by

, , , ,jt j ul sl sh= can be shown to satisfy the following recursive equations

10

( ) 1 11ul l l ult t t tt t tul b q Ew η + += − − − Λ+ , (23) ( ) ( ) ( )1 1 1 1 11 1sl l l sl l h sh slt t t t t t t t tt tl ts b q E e Ew qη η+ + + + += − − − Λ − Λ −+ + , (24) ( ) 1 11sh h h sht t t tt t tsh b q Ew η + += − − − Λ+ . (25) Assuming that the bargaining power of workers is job-type specific, and denoting it by

, ,,j j h lα = the corresponding Nash bargaining functions are ( ) ( )1 ,l lul ult tα α−

( ) ( )1 ,l lsl slt tα α−

and ( ) ( )1 ,h hsh sht tα α−

which imply the surplus sharing rules

( )1 0,ul ul

ul ult tt ll l tul u

t tw wα α∂ ∂+ − =

∂ ∂

( )1 0sl sl

sl slt tt tsl sll

tl

tw wα α∂ ∂+ − =

∂ ∂

and

( )1 0.sh sh

sh sht tt hh h tsh s

t tw wα α∂ ∂+ − =

∂ ∂

Together with equations (16), (17) and (22) to (25),

these yield the following solutions for ,ul slt tw w and shtw which have the standard interpretation:

( )1 1 1 ,l

ul l l l ul l

ltt t t t t t tul

t

Hw w f q E bN

α αη + +∂

= +

− + Λ −∂

(26)

( ) ( )1 1 1 11 1 ,l

sl l l l sl l h shtt t t t t t t t t tl l t

tl s

Hw w f q E eq E bN

α η η α+ + + +∂

= − + Λ∂

−

Λ

− − + (27)

( )1 1 1 .h

sh h h h sh h

htt t t t t t tsh

t

Hw w f q E bN

α αη + +∂

= +

− + Λ −∂

(28)

2.4. The final good sector The final good sector (the downstream industry) produces a homogenous good competitively using domestically produced (in the upstream industry) and imported varieties of an intermediate good with a CES technology

( ) ( )*

11 1 1 1/1 1/ 1 1/* * dt it it

i M i M

Y M y di M y diσσ σ

σ σ−− −− −

∈ ∈

= +

∫ ∫ , (29)

where Y is the quantity of the final good, dity and *ity and M and *M are, respectively, the

quantities and the masses of domestically produced and imported intermediate input varieties.

Denoting the output and input prices respectively by *and and ,dt it itP p p the sector’s profit is

*

* *d dYt t t it it t it it

i M i M

PY p y di p y diτ∈ ∈

Π = − −∫ ∫ where τ is the iceberg trade cost. Maximisation of profit

yields the demand functions

, ,d

d t itit

t

Y py i MM P

σ−

= ∈

(30)

11

*

* ** , .

t t itit

t

Y py i MM P

στ

−

= ∈

(31)

The price index dual to (29),

( ) ( )*

11

1 1**

1 1dt it it

i M Mt

i

P p di p diM M

σσ σ

τ−

− −

∈ ∈

= +

∫ ∫ , (32)

then ensures that 0YtΠ = .

2.5. The intermediate good sector The mass M of intermediate varieties is assumed to be time-invariant and each variety is produced by a firm whose total factor productivity is denoted by ϕ. We assume that firms differ

in their productivity and that ϕ is distributed over the [1, )∞ support with a time-invariant density function. We therefore use the productivity parameter to distinguish between firms and firm-level variables and simplify notation by dropping the variety index, .i M∈ Denoting export-related variables by superscript x, a typical firm’s input requirement for its domestic and export production is given by

( ) ( ) ( ) ( ),d d x xt t t ty a y aϕ ϕ ϕ ϕ ϕτ ϕ= = , (33) where ( ) ( ) ( )( )d xt t ta a aϕ ϕ ϕ≡ + is a composite input consisting of capital, k, and labour hl and

ll respectively employed in high-tech and low-tech jobs. We assume that these primary factors are combined by the Cobb-Douglas technology,

( ) ( ) ( ) ( )h lk h l

t t tt

k h l

k l la

ϑ ϑϑϕ ϕ ϕ

ϕϑ ϑ ϑ

=

, 1k h lϑ ϑ ϑ+ + = . (34)

A firm’s cost of production therefore is

( ) ( ) ( ) ( )a h h l lt t t t t t t tp a rk lw w lϕ ϕ ϕ ϕ= + + , (35)

where ap is the unit price of a. Efficiency requires

( ) ( )ak tt t tr k p aϑϕ ϕ= , (36)

( ) ( )th h at t h tw l p aϑϕ ϕ= , (37) ( ) ( )tl l at t l tw l p aϑϕ ϕ= , (38)

( ) ( ) ( )h lka h lt tt tp r w wϑ ϑϑ= . (39)

For domestic sales, the firm’s real profit is ( ) ( ) ( ) ( )d d d a dt t t t t tp y P p aπ ϕ ϕ ϕ ϕ= − which is

maximised subject to ( ) ( ) ( )( )d dt t t ty Y M p Pσ

ϕ ϕ−

= , yielding

( )( )1

d at t

t

p pP

ϕ σσ ϕ

=−

. (40)

12

The real profit from exporting is ( ) ( ) ( ) ( )x x x a xt t t t t tp y P p aπ ϕ ϕ ϕ ϕ= − . We assume, for simplicity, that the foreign demand for a typical domestic variety is

( ) ( ) ( )( )* *x xt t t ty F M p Pσ

ϕ ϕ−

= where *P and *F are the relevant foreign price level and the

scale factor representing the real foreign income share spent on the good. It can be verified

that ( ) ( )x dt tp pϕ τ ϕ= maximises ( )xtπ ϕ . 2.6. General equilibrium Based on the above results, the following relationships hold for any two productivity values, e.g. ϕ and ϕ :

( )( )

( )( )

( )( )

( )( )

1 1

; ; ; ; , .j j j j

t t t tj j j j

t t t t

p y aj d x

p y a

σ σ σϕ ϕ ϕ π ϕϕ ϕ ϕ ϕϕ ϕ ϕ ϕ ϕ ϕ π ϕ ϕ

− − −

= = = = =

(41)

Defining the average industry productivity as in Melitz (2003) and hence setting

( )1

11g d

σσ

ϕ

ϕ ϕ ϕ ϕ−

−

= ∫ , (42)

we can express all aggregate measures in terms of ϕ – e.g. the aggregate demand for capital

by all firms is ( ) ( ) ( ) ( ) ( )D d x d x d xt it it t t t ti M

K k k di Mg k k d M k kϕ

ϕ ϕ ϕ ϕ ϕ ϕ∈

= + = + += ∫ ∫ .

The labour supply functions are defined as concave relationships that convert the number of a given type of worker to the respective man-hours. For the low-tech and high-tech jobs, we assume

( ) ( ) ,ul sll ul slul slt t tul sl

h hH N Nψ ψ

ψ ψ= + (43)

( ) ,shh shsht tsh

hH Nψ

ψ= (44)

where jh and , , , ,j j ul sl shψ = are constant positive coefficients and the latter captures the

required decreasing returns to scale. The respective man-hour wage rates paid by the firm, ltw

and ,htw are then determined by the labour market clearing condition that equates demand and

supply for man-hours,

( )l lt tMl Hϕ = , (45) ( )h ht tMl Hϕ = . (46) The government operates a balanced budget and finances unemployment benefit and public investment gI with revenues generated through lump-sum tax from households and firing fees

13

from employment agencies as well as the profit of the latter which we assume to be owned publicly. Thus,

( ) ( ) ( ), , .,g ul l h tsl sh l sl l ht sh hult t t t t t t t t tb S I T N N f N f N V VN Nη π πη + ++ = + + + (47) In addition to consumption and private and public investments, the demand for the final good, i.e. domestic absorption denoted by Y, includes the vacancy creation costs and is given by g l l h ht t t t t t t tY C I I c V c V= + + + + ,

and GDP is the sum of domestic absorption and its foreign equivalent (net exports), hence,

( ) ( )* * *d x

t t t t t tt

t

M p y M p yD

PG P Y

τ ϕ ϕ τ=

−+ , (48)

where we have assumed * *jt tp p= and * *it ty y= .

We assume that capital mobility is governed by the (exogenously determined) rule

( )* Dt t t t tr r K K Kκ− = − , (49) where the excess demand for capital raises r above *r by an amount determined by the given

value of 0κ ≥ which is an inverse measure of capital mobility: 0κ = corresponds to perfect mobility and the country can, in principle, sustain any excess demand for capital at the rate

*r r= ; κ → ∞ instead frees any ties between r and *r and requires it to be determined by the

capital market clearing condition DK K= which ought to hold in this case. The balance of payments, which requires the value of net exports to match the interest payments on net capital flow,

( ) ( ) ( )* * *

*d x

t t t t t tDt t t

t

M p y M p yr K K

Pτ ϕ ϕ τ−

− = , (50)

can be shown to hold as long as all markets are cleared. In order to obtain closed form solutions, we assume that firms’ productivity parameter has a time-invariant Pareto distribution,

( ) [ )(1 ) , 1, , 1g γϕ γϕ ϕ γ σ− += ∈ ∞ > − . (51) Thus, equation (42) implies

( )

11

1

σγϕγ σ

− = − −

. (52)

We also assume that the household utility function has the following functional form

( ) ( ) ( )1 1 11

1 1 1 1

u sl shc

u u sl sl sh sht t tt

tc u sl sh

A X A X A XCUυ υ υ

υ

υ υ υ υ

+ + +−

= − − −− + + +

, (53)

14

where 1 cυ is the intertemporal elasticity of substitution and jA and , , ,j j u sl shυ = , are

constant positive parameters respectively capturing the weight and elasticity attached to the disutility of participation in the labour force. Finally, we assume a standard Cobb-Douglas constant returns to scale technology for matching

( ) ( ) ( )1g

l ll l tm Kl ul sl lt t t te S S V

λ λε −+= + , (54)

( ) ( )( ) ( )11g h hh h tm Kh sh l sl h

t t t te S e N Vλ λε η

−+= + − , (55)

where, for , ,j h l= ( )0,1jλ ∈ are match elasticities and gj j tm Kε+ are job-specific measures

of matching efficiency where 0, 0j jm ε> ≥ , and ( ) 11g g g gt t tK K Iδ −= − + is the public capital in employment services that enhance matching efficiency.8 In essence, jm can be thought of

as the underlying level of matching efficiency of a given segment of the labour market. For a given jm , the effective quality of matching depends on the investment in employment services,

gtI . This form of public investment in labour market services is seen as a cost-effective way to

reduce the frictions that characterise the market and has featured in labour market policies in many countries in recent years. Particularly, this can include the adoption of information technology that influences the way jobs are applied for and/or advertised by firms which reduces search times and information asymmetry, or a form of investment in upgrading public job centres.9 The effect of a rise in gtI is an important but scarcely studied instrument of labour

market reform which enables policy interventions to target the efficiency of job search. 3. Calibration We calibrate the model’s steady state to reflect the stylised characteristics of the UK economy, with emphasis on the labour market features. We assume a quarterly time-frequency and base the calibration of all parameters on empirical evidence and data averages. When these are lacking, we use the values commonly used in the relevant literature to maintain comparability. In particular, following the common practice in the literature, we use the standard values for the subjective discount factor and capital depreciation rate, 0.99β = and 0.025δ = , and

8 As documented by Petrongolo and Pissarides (2001), the matching between job searchers and vacancies explains roughly 50% of the labour market flows (of jobs and of workers between activity and inactivity). 9 In Germany, the restructuring of the federal employment agency, as part of the Hartz reform between 2003 and 2005, was aimed at improving job matching efficiency (Krebs and Scheffel, 2013) and was found to explain about 23% of the decrease in unemployment in the following years (Launov and Wälde, 2016). In the UK, the complete overhaul of the Jobcentre Plus resulted in the introduction of Jobseeker Direct (a telephone job matching service) (Riley et al., 2011) and the Universal Job Match Service (offering a comprehensive ‘one-stop-shop’ for the unemployed allowing them to upload CVs and apply online within the same platform (European Commission, 2017). Mosseri-Marlio (2016) argues that digital tools, relying on data driven intervention, can drastically improve job centres’ effectiveness.

15

normalise the elasticity parameters in the utility function by setting 1, , , , .j j c ul sl shυ = = We

also normalise GDP to unity and assume that trade is balanced in the initial steady state equilibrium. OECD (2016, 2018) data show that the ratio of unskilled and skilled workers are respectively 0.56 and 0.44 for the UK, where the former group is identified as those with at most an upper secondary education. These values enables us to target inactivity rates of the

skilled and unskilled, 0.12sL Ζ = and ( )1 0.28,uL − Ζ = based on the empirical evidence provided by Gomes (2012). The average share in the labour force of employed and unemployed in the UK between 2008 and 2014 are respectively 73% and 6%, based on ONS statistics. These imply an aggregate inactivity level of 21% given the normalisation of the household population to unity. Using these values and allowing for the scale parameters for the disutility of labour

market participation , , , ,jA j u sl sh= to be freely determined by the model, we target aggregate

unemployment rate ( ) ( )ul s u su S S X X= + + within the 5% – 8.4% range, to match the figures reported by the OECD statistics for the UK over the 2008-2015 period. The measure of OTJS

efficiency is set as e = 0.1 to yield the mismatched employment ratio ( )sl ul sl shN N N N+ + within the 0.13–0.15 interval as observed in the UK (ONS, 2016).

Job destruction rates, hη and lη , are respectively set to 0.009 and 0.02 based on the

empirical estimates reported in Gomes (2012). The initial steady state unemployment benefit payment is set so that the corresponding benefit replacement rate, based on evidence provided

by van Vliet and Caminada (2012), is ( ) ( ) 0.23ul sl sh ul ul sl sl sh shb N N N N w w N w N+ + + + = . We also assume symmetric bargaining across the job spectrum, set 0.5ul sl shα α α= = = and

impose the Hosios (1990) efficiency condition which implies 0.5l hλ λ= = .

Assuming that trade and capital mobility are both frictionless and free to start with, we set

1τ = and 0κ = in the benchmark calibration. The latter implies *r r= . Using the foreign final good as the numeraire, we normalise its price to unity setting * 1P = . Utilising the relevant trade-related series over the 2008-2014 period from the World Bank Development Indicators dataset (WDI, 2016), we calculate the scale factor in foreign demand and the relative price of

exported to foreign varieties respectively as * *0.415 and 0.785xF p p= = and, to ensure that

our calibration reflects the actual UK to world GDP ratio, we set * 0.0465M M = .

Given that we start with a balanced trade, DK K= holds initially and is sustained by private investment, which is set consistently with the UK investment/GDP ratio of 16.61% over the period 2008-2014. Using the data from EU-KLEMS (2016), we set the labour input

elasticities 0.44hϑ = and 0.26lϑ = respectively, corresponding to the average values over the

16

2008-2015 period, and let 1k h lϑ ϑ ϑ= − − for consistency with the constant returns to scale.10 The values of the elasticity of substitution and the shape parameter of the Pareto distribution of firms’ productivity, σ and γ , are set to yield a profit/output ratio of roughly 20%, corresponding to the average UK business profit share over the 2008-2014 period.11 The chosen values, 4.5σ = and 3.8γ = are within the range used in similar studies and satisfy

1γ σ> − .

The values chosen for vacancy creation unit costs and labour input conversion technologies are 0.308lc = and 3.628hc = , 5, 0.06 0.06ul slh h= = and 0.111.shh = These are set to yield

the steady state wage ratio 0.62l hw w = , consistent with the average wage ratio of non-graduates to graduates reported over the 2008-2016 period (Department for Education, 2017). In order to allow for sufficient concavity, in converting labour to man-hours in equations (43)

and (44), we follow Christoffel and Kuester (2009) and set 0.995ul sl shψ ψ ψ= = = . Finally, to

explore the quantitative effects of allowing government investment in matching efficiency, we set 0.003gI GDP = which reflects the UK’s GDP share of public expenditure on Employment

Services (ILO, 2015) and let 0.009gδ = , which corresponds to the ratio of private to public

capital depreciation rate of 0.36 as reported in Angelopoulos et al. (2012). We also use 0.15l hε ε= = and choose the values of lm and hm consistently with the value of mismatch of

roughly 0.13. Our benchmark solution for the immediately relevant variables, corresponding to the calibration described above, is given in column 2 of Table 1 in the Appendix and was found to be robust to sensitivity analysis in which we perturbed the values of parameters and exogenous variables.

4. De-globalisation We first examine the effects of trade frictions with a central focus on labour market outcomes. Trade flows on the whole have slowed down worldwide after the 2007-2008 financial crisis. At the same time, revival of protectionist stances and backlashes against globalisation have also resulted in political developments that are likely to raise trade barriers and reduce international trade further – e.g., Brexit in the UK, the rise of anti-EU sentiments in other EU nations, and the recent trade policies of the Trump administration in the US. To examine the impact of raising trade frictions, we consider a one-off permanent increase in iceberg trade cost, τ , e.g., as might results from a ‘hard Brexit’. Column 3 of Table 1 in the Appendix reports the resulting solution. The immediate effect of a higher τ is to increase the effective prices of both exported and imported intermediate varieties, leading to a reduction in

10 Source: http://www.euklems.net. 11 Sources: http://ec.europa.eu/eurostat/web/sector-accounts/data/ annual-data.

17

the volume of trade as, ceteris paribus, both foreign demand for domestically produced varieties and domestic demand for foreign varieties drop. The higher import price also raises the production cost in the final good sector as the intermediate price level rises. The latter, together with the higher export prices, reduces the level of aggregate demand in both downstream and upstream sectors. Despite leading to a substitution away from foreign towards domestic varieties and consequently inducing some shifting of resources from exports to domestic production, the reduced aggregate demand resulting from a higher trade cost leads to a drop in demand for factors by firms. This lowers vacancy creation incentives and, consequently, job finding probabilities – reflected in lower employment across all job types. To understand the impact of trade frictions on match quality, we utilise two measures. First, we define the mismatch ratio as

( )( )1 ,sl l h sl st t t tMMR X e q N Xη= − − (56) where the numerator corrects the mass of skilled workers opting for low-tech jobs to account for those whose OTJS for a high-tech job has been successful. Thus, all else equal, the lower is the mass of skilled workers that are willing to participate in mismatch employment, the higher is the quality of match which will further improve the higher is the success of OTJS. The second measure of match quality is a modified version of the aggregate skill dispersion indicator recommended by Kiss and Vandeplas (2015), namely

( )1sh ul sl

sh sl ul sh sl ul

N N NASDI Z ZN N N N N N

+= − + − −

+ + + +, (57)

which accounts for the actual size of employment in the low-tech task jobs. As indicated by our results, the deterioration in labour market prospects leads to lower match quality – as reflected by increases in both measures of skill mismatch – and underpins the shrinking of the labour force. Despite this, competition for jobs increases and so do unemployment rates. Interestingly, the downward pressure on wages that results from higher trade frictions provides the only recovery channel since markup pricing of the intermediate goods leads to a reduction in the price of domestic varieties. However, this channel is not sufficiently strong to overcome the negative impact of the rise in trade cost: domestic and foreign absorption both fall reducing GDP; labour force and employment both shrink, and the level of economic activity drops. Broadly, these results match observed empirical regularities. For instance: Barattieri et al. (2018) find that protectionist shocks are recessionary; Cooray et al. (2017) document empirically the adverse effect of trade frictions on the size of labour force; Felbermayr et al. (2011b) discuss the greater unemployment consequences of trade restrictions; and Davidson et al. (2014) and Krishna et al. (2014) also offer evidence of improved firm-worker sorting as a result of trade integration. Finally, it might be argued that as far as employment is concerned unskilled workers tend to ‘benefit’ from trade barriers as it improve their ‘relative position’,

with ul shw w rising and ul shS S falling as τ increases. But the economy clearly experiences a

18

higher incidence of skill mismatch as both MMR and ASDI deteriorate and in fact ul slw w falls

whilst ul slS S rises. This is consistent with empirical evidence that shows that falling trade barriers lead firms to change workforce structure by increasing firm-level shares of high-skill occupations and result in a lower degree of mismatch between workers’ skills and those demanded by firms (e.g. Davidson et al. 2014). 5. Reforming the Labour Market Turning to labour market reform, we start from the benchmark calibration which portrays a liberal welfare state system and examine how labour market policies affect the economy when they are implemented, first individually and then together as a flexicurity reform package, hereafter referred to as FRP. As previously noted, from a methodological view point, since changes in individual policies may have opposite effects on the equilibrium values of the variables, the extent to which they are altered relative to each other is an important determinant of the net impact of a given reform package. To this end, we use Denmark as our example of the flexicurity system and change the relevant UK policy parameters in the direction of and by a proportion consistent with taking them to their corresponding Danish counterparts. In particular, FRP entails: (i) increasing the unemployment benefit rate b by 60% (based on the estimates provided by Nickell et al., 2005; Vliet and Caminada, 2012); (ii) raising the firing

cost f by 43% (OECD, 2013); and (iii) reducing the unit vacancy creation costs lc and hc by 50% (The World Bank, 2016). The corresponding solutions for our variables of interest can be found in columns 4, 5 and 6 of Table 1. The increase in b, akin to raising job seeker allowance, raises the opportunity cost of leisure and workers’ outside option in wage negotiations. These respectively lead to an increase in job search activities and higher bargained wage rates. The latter work towards reducing the profitability of job matches and the number of new vacancy created. Since the matching function is increasing in both vacancies and searchers, the impact of a rise in b on aggregate job matches depends on which effect dominates. With exogenous participation (e.g., as in Cacciatore et al., 2016), job matches would inevitably reduce due to the adverse effect of b on job creation. However, with endogenous participation, the resultant flow out of non-market activities enlarges the pool of job seekers, implying that the adverse effects of a shock on the aggregate number of matches can potentially be reversed by an increase in participation. Our experiments reveal that the effect of a higher b on the number of job searchers and vacancies cancel out, leaving aggregate job matches approximately unchanged.12 This effect, combined with higher wages, results in a new equilibrium characterised by a higher level of economic activity as GDP, domestic and foreign absorption all rise. There is however a moderate

12 The aggregate labour market response to an increase in b resonates the evidence in Bruckner and Pappa (2012) who find that a fiscal shock that raises aggregate demand can result in both higher employment and unemployment by inducing greater participation which reinforces the number of job searchers.

19

decrease in welfare due to the enlarged participation. As for the impact of raising b on match quality, we find both mismatch measures to deteriorate. The reason for this is that, by encouraging greater participation across the skill spectrum and reducing vacancy creation activities, a rise in b results in tighter job competition that leads to an increase in the number of skilled workers willing to accept low-tech jobs with adverse effect on match quality.13 Moreover, the decrease in high-tech job finding rate negatively affects the success of OTJS, further worsening the quality of job matches. Thus, the endogeneity of workers’ participation in the labour market is a key channel in the transmission mechanism of the policy and implies that a passive labour market policy such as unemployment benefits can perform as an activation measure. Next, we consider the effect of an increase in f which can be interpreted as raising the level of employment protection. The immediate impact of this is to reduce the value of employment and profits from job matches. Consequently, incentives for vacancy creation decline, resulting in a lower employment. As one would expect, the contractionary effect of raising employment protection reduces the level of economic activity, with the new equilibrium being characterised by lower levels of employment, GDP and domestic and foreign absorption and by a higher level and rate of unemployment. Perhaps surprisingly, skill mismatch tends to improve when f is raised. The main reason for this lies in the adjustment of the participation margin: as market tightness falls, so do searchers’ job finding rates, resulting in higher unemployment duration and in an outflow from the labour force which mitigates the higher competition for jobs. Overall, however, we find the quantitative effects of this policy to be rather small relative to the other polices (which is consistent with the ambiguous effects of employment protection found in the empirical literature).

Finally, we examine the impact of reducing lc and hc . The immediate effect of lowering vacancy creation costs is to increase market tightness, hence raising the number of job matches, employment and bargained wages for all searchers. The latter stimulates participation and the resultant improvements in job-finding probabilities, together with increase in OTJS transition rate, leads to an improvement in match quality – reflected in decreases in MMR and ASDI. Consistently, the rate of unemployment across all workers also falls. The resulting boost in aggregate demand leads to a higher level of economic activity; GDP and both domestic and foreign absorption rise, and the welfare level also improves. In the last column of Table 1 we show the combined effect of the above policies, namely our proposed FRP. As it can be seen, the stimulating effect of the individual polices dominate and the reform package as a whole leads to an increase in the levels of participation, employment and GDP and to a reduction in the level and rate of unemployment. Both domestic and foreign absorption increase, and the economy enjoys an improvement in skill mismatch 13 The effect of increases in unemployment benefits on match quality here contrasts that of Lisa et al. (2016) who argue that unemployment benefit could lead to improved match quality because it induces workers to be picky about the match type.

20

and welfare level. These results are broadly consistent with observed empirical regularities. For instance, European Commission (2013) finds that countries with low expenditures on active labour market programmes experience greater skill mismatch. In an earlier study, Lehman and Kluve (2010) had already come to a similar conclusion, arguing that job creation subsidies can result in a higher matching efficiency that could lead lower job mismatch.

6. Post-Reform Trade De-Globalisation In this section we examine the extent to which the effects of higher trade restrictions could be influenced by a possible FRP. To this end, we start from the free and frictionless trade and capital mobility benchmark calibration, implement the FRP, and then gradually increase the trade cost τ , e.g. to mimic the effect of scenarios characterising the “Soft” and “Hard” Brexit debate – as an example, see Van Reenen (2016). We also examine the effect of combining the implementation of FRP with either of the following: (i) an increase in public expenditure

targeting matching efficiency (a rise in gI ); (ii) an increase in the density of low productivity firms (a rise in the shape parameter of the firm-level productivity distribution γ );14, and (iii)

introducing some friction in the cross border mobility of capital – by letting 0.κ > 15 Our results, for a selected subset of variables, are displayed in the figures presented in Table 2 in the Appendix. Each figure depicts the pre- and post-ERP response of a variable to rising trade cost τ, compared with the corresponding benchmark pre-reform solution at τ = 1. Before

examining the scenarios which combine the FRP with changes of ,gI γ or κ, we note that on the whole the graphs confirm that if implemented prior to introducing trade frictions, the FRP puts the economy in a stronger position to withstand the adverse consequences of raising trade barriers. As shown by the results, the effects of trade costs are similar qualitatively in both pre- and post-reform regimes. However, the trade cost ought to reach a certain level before the positive effects of the FRP are fully eroded. For instance, the post-FRP value of GDP does not fall back to its pre-reform benchmark value until τ is raised by about 10%; in the absence of the reform, such increase in trade cost would results in approximately 3% decrease in GDP. The initial effects of combining the FRP with a policy aimed at enhancing the efficiency of job-matching can also be seen, in the figures in Table 2 in the Appendix, by comparing the corresponding post-reform solution values at 1τ = ; these show the impact of raising gI , from

its benchmark value of 0.003 to 0.006, an increase that is commensurate to the difference between the UK and Danish levels of public expenditure on Employment Services. Augmenting the FRP with this additional policy strengthens its impact: by facilitating job matches via a simultaneously increasing job finding and vacancy filling probabilities, the latter

14 A number of studies have highlighted the importance of firm-level size and productivity in explaining changes in aggregate labour market performance (e.g., Elsby and Michaels, 2013; and Görg et al., 2017). 15 De-globalisation can also involve a reduction in the degree of capital mobility (e.g., Greenaway and Nelson, 2001; Antràs and Caballero 2009) and influence labour market outcomes (e.g. Vallanti, 2015).

21

results in a lower level of unemployment across all worker categories as well as a reduction in the duration spell of both unemployment and vacancies. Consistent with Riley et al. (2011), who evaluate the impact of job-brokering on labour market outcomes, our results show that a higher investment in matching efficiency leads to a lower aggregate level of inactivity. The mass of skilled participation in mismatched employment however rises, but the ensuing increase in job-to-job flow leads to improvement in job-match quality. While augmenting the

FRP with an increase in gI does not alter the qualitative effect of higher trade costs,

quantitatively it increases the threshold level of trade cost which neutralises the positive impact of the reforms relative to the benchmark. More generally, a higher investment in employment services is far more effective in sustaining participation, low rates of unemployment and high rates of OTJS and, consequently, greater job match quality as trade cost increases. How does a shift in the composition of firms change the impact of the FRP? The figures in Table 2 show this by comparing the post-FRP solutions corresponding to 3.8γ = (the value

used in the benchmark scenario) and to 3.825.γ = Since an increase in γ implies a higher the density of firms with lower productivity, the profitability of job matches and vacancy creation both fall and, as a result, the effectiveness of the FRP diminishes. This change is therefore equivalent to enhancing the effect of rising trade costs – which is particularly evident in the

case of GDP, P, ,dp unskilled and mismatched unemployment rates and MMR – as, on the whole, benefits of the FRP are potentially erased at much lower levels of trade cost. Finally, we consider how introducing some degree of capital mobility friction alters the effectiveness of the FRP. In a theoretical model such as ours, the extent of capital mobility, characterised by the response of capital flows to interest rate differentials, enables the economy to accommodate an excess demand or supply of capital that is consistent with the trade balance. Specifically, with capital mobility frictions, the interest parity can no longer be attained and the interest rate differential *r r− will be determined by the discrepancy between the domestic firms’ demand for the stock of capital and households’ accumulated capital stock which ought to be met by capital flow. Consequently, the balance of payment will only hold if the resulting interest payments on capital inflow (outflow) is matched by a trade deficit (surplus). Put differently, the economy can sustain a trade deficit or surplus as long as it is offset by the return on capital flows; the higher is the barriers to capital flows, the smaller is the sustainable magnitude of the trade deficit/surplus. Thus, the impact of raising capital mobility frictions on the economy is likely to be contingent on whether an economy is initially in a position of trade surplus or deficit. Starting from a trade surplus position where the economy is a net exporter, the overall effect of an increase in such frictions will be contractionary. The opposite would hold if the economy were initially a net importer. To illustrate this, we also plot in the figures in Table 2 the post-FRP solutions corresponding to 0.25,κ = which could be compared to those with 0κ = (the value used in the benchmark scenario). The impact of this on trade is

22

evident in the figure which plots ‘Net Exports’ where the trade surplus achieved by implementing the FRP at 1,τ = is dramatically reduced by raising κ. However, since restricting capital mobility restrains the size of the trade balance, it also moderates the adverse effects of rising trade costs after the latter has led to a trade deficit. This is clearly depicted in the GDP figure: a higher κ dampens the negative effect of raising τ after the latter exceeds the trade balance threshold. As far as labour market outcomes are concerned, the impact of restricting capital mobility is more evident in the case of labour force participation and employment where the direction of the impact is altered. More specifically, raising τ stimulates participation and increases employment which contrasts with what happens when 0κ = . The underlying intuition, however, is straightforward. The stock of capital used by firms, as a primary factor of production, is substitutable by labour. When capital mobility is frictionless, firms enjoy an infinitely elastic supply of capital at a constant rate *r r= . Imposing capital mobility frictions change this such that excess demand for capital raises r above *r and induces factor substitution. Although capital mobility frictions reduce the employment level of mismatched workers as trade cost increase, the aggregate employment effect is dominated by the skilled and unskilled employment which tend to increase as τ rises.

7. Summary and Conclusions We have examined the employment consequences of reducing the level of economic integration (as it might, for instance, result from Brexit) of an open economy characterised by vertical linkages in production, imperfect goods, skilled and unskilled labour markets. We find that raising trade barriers leads to under-utilisation and misallocation of resources and results in higher unemployment rates across skill levels, a lower job-match quality and a reduced level of economic activity and income. These effects are enhanced if the firm-level productivity distribution is more skewed towards less productive firms. Maintaining frictionless cross-border capital flows does not necessarily moderate the negative effects of raising trade costs. We also find that implementing a reform package which moves an economy with a liberal welfare state system in the direction of flexicurity – entailing lower degrees of labour market flexibility and greater unemployment insurance – enables it to better withstand the adverse effects of increasing trade costs. The positive effects of such a reform are further amplified if the package is augmented with a higher level of public investment in job-search/matching efficiency. A broad implication of the paper is that an efficient labour market does not need to be thin on worker security. Crucial is not ‘labour market flexibility’, but the ‘social investment’ dimension of the WS: support for the unemployed, social investment and activation policies. Importantly, unemployment insurance can act as an activation policy by fostering labour market participation, an effect that is strengthened if coordinated with other active labour market polices.

23

References

Adalet McGowan, M. and D. Andrews (2015a), Labour Market Mismatch and Labour Productivity: Evidence from PIAAC Data, OECD Economics Department Working Papers, No. 1209, OECD Publishing, Paris. http://dx.doi.org/10.1787/5js1pzx1r2kb-en.

Adalet McGowan, M. and D. Andrews (2015b), Labour Market Mismatch and Labour Productivity: Evidence from PIAAC Data, OECD Economics Department Working Papers, No. 1209, OECD Publishing, Paris. http://dx.doi.org/10.1787/5js1pzx1r2kb-en.

Albrecht, J. and S. Vroman (2002). A Matching Model with Endogenous Skill Requirements, International Economic Review, 43, 283–305.

Angelopoulos, K., J. Malley and A. Philippopoulos (2012). Tax structure, growth, and Welfare in the UK, Oxford Economic Papers, 64, 237–258.

Antràs, P. and R. J. Caballero, (2009). Trade and Capital Flows: A Financial Frictions Perspective, Journal of Political Economy, 117, 701-744.

Arseneau D.M. and B. Epstein (2017). Offshoring, Mismatch, and Labor Market Outcomes. In: Christensen B., Kowalczyk C. (eds) Globalization. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-49502-5_3.

Arseneau, D. M. and B. Epstein (2014). The Welfare Costs of Skill-Mismatch Employment, Finance and Economics: Board of Governors of the Federal Reserve System, Discussion Series 2014–42. https://www.federalreserve.gov/pubs/feds/2014/201442/201442pap.pdf.

Autor, D. H., D. Dorn and G. H. Hanson, (2013). The China Syndrome: Local Labor Market Effects of Import Competition in the United States, American Economic Review, 103, 2121–2168.

Azariadis, C. and C. Pissarides, (2007) Unemployment Dynamics with International Capital Mobility, European Economic Review, 51, 27-48.

Barattieri A., M. Cacciatore, and F. Ghironi (2018). Protectionism and the Business Cycle, NBER Working Paper 24353.

Brunello, G. P. Garibaldi and E. Wasmer, (2007), Education and Training in Europe, (eds), Oxford University Press.

Cacciatore, M. (2014). International Trade and Macroeconomic Dynamics with Labor Market Frictions Journal of International Economics, 93, 17–30.

Cacciatore, M., G. Fiori and F. Ghironi (2016), Market deregulation and optimal monetary policy in a monetary union, Journal of International Economics, 99, 120-137.

Cooray, A., N Dutta, and S. Mallick (2017). Trade Openness and Labor Force Participation in Africa: The Role of Political Institutions, Industrial Relations, 56, 319-350.

Coşar, A. K., N. Guner and J. Tybout (2016). Firm Dynamics, Job Turnover, and Wage Distributions in an Open Economy, American Economic Review, 106, 625–663.

Davidson, C., Heyman, F., Matusz, S., Sjöholm, F. and S. Chun Zhu, (2014), Globalization and Imperfect Labor Market Sorting, Journal of International Economics, 94, 177–194.Davidson, C., Matusz, S., 2012. A Model Of Globalization and Firm-Worker Matching: How Good is Good Enough?, Int. Rev. Econ. Finance. 23, 5–15.

Davidson, C., S., Matusz and A. Shevchenko, (2008), Globalization and Firm Level Adjustment with Imperfect Labor Markets, Journal of International Economics, 75, 295–309.

Davies, B. (2018), “Help not Hassle: A Case for Better Employment Support”, forthcoming, IPPR Progressive Review.

Draghi (2014), Unemployment in the Euro Area, Speech by Mr Mario Draghi, President of the European Central Bank, at the Federal Reserve Bank of Kansas City Economic Symposium, Jackson Hole, Wyoming, European Central Bank.

Department for Education, (2017), Graduate Labour Market Statistics 2016.

http://dx.doi.org/10.1787/5js1pzx1r2kb-enhttp://dx.doi.org/10.1787/5js1pzx1r2kb-enhttps://doi.org/10.1007/978-3-662-49502-5_3https://www.federalreserve.gov/pubs/feds/2014/201442/201442pap.pdfhttps://www.ecb.europa.eu/press/key/%20date/2014/html/sp140822.en.htmlhttps://www.gov.uk/government/uploads/system/uploads/attachment_data/file/610805/GLMS_2016_v2.pdf

24

ECB (2012), Euro Area Labour Markets and the Crisis, Structural Issues Report, European Central Bank.

Elsby, M.W.L. and R. Michaels (2013). Marginal jobs, heterogeneous firms, and unemployment flows. American Economic Journal, 5, 1-48.

Elsby, M.W.L, B. Hobijn and A. Şahinc, (2015). On the importance of the participation margin for labor market fluctuations, Journal of Monetary Economics, 72, 64-82.

European Commission (2017). Universal Jobmatch – Online Vacancy Platform for Employers and Jobseekers.

European Commission (2013), Employment and Social Developments in Europe 2012: The Skill Mismatch Challenge in Europe, EU Commission Staff Working Papers, No. 9.

EU-KLEMS (2016). Labour Input Data: United Kingdom. Felbermayr, G., J. Prat and H. Schmerer (2011a). Globalization and Labor Market outcomes:

Wage Bargaining, Search Frictions, and Firm Heterogeneity, Journal of Economic Theory, 146, 39–73.

Felbermayr, G., J. Prat and H. Schmerer, (2011b), Trade and unemployment: What do the data say? European Economic Review, 55, 741–758.

Gaddis, I. and J. Pieters, (2012). Trade Liberalization and Female Labor Force Participation: Evidence from Brazil, IZA Discussion Paper No. 6809.

Ghironi F., and M. Melitz (2005). International trade and macroeconomic dynamics withheterogeneous firms, Quarterly Journal of Economics, 120, 865-915.

Greenaway, D. and D. Nelson, (2001). Globalisation and Labour Markets: Literature Review and Synthesis, Globalisation and Economic Policy Research Paper 2001/29, Nottingham.

Green, F., S. McIntosh, and A. Vignoles (1999). Overeducation and Skills: Clarifying the Concepts, Centre for Economic Performance, London School of Economics and Political Science, CEPDP, 435.

Gomes P. (2012). Labour market flows: Facts from the United Kingdom, Labour Economics, 19, 165–175.

Görg, H., P. Henze, V. Jienwatcharamongkhol, D. Kopasker, H. Molana, C. Montagna, F. Sjöholm (2017). Firm Size Distribution and Employment Fluctuations: Theory and Evidence, Research in Economics, Special Issue in Honour of 40 Years of the Dixit-Stiglitz Model, 71, 690-703.

Helpman, E., O. Itskhoki and S. Redding (2010), Inequality and Unemployment in A Global Economy, Econometrica, 78, 1239–1283.

Helpman, E., O. Itskhoki (2010). Labour Market Rigidities, Trade and Unemployment Review of Economic Studies, 77, 1100–1137.

Hosios, A. J. (1990). On the Efficiency of Matching and Related Models of Search and Unemployment, Review of Economic Studies, 57, 279–298.

ILO (2014). Skill Mismatch in Europe, Statistics Brief, International Labour Office, Geneva. Kiss A. and A. Vandeplas (2015). Employment, Social Affairs & Inclusion: Measuring skills

mismatch, the European Commission Analytical Web Note 7/2015. Krebs, T and M. Scheffel (2013). Macroeconomic Evaluation of Labor Market Reform in

Germany, IMF Working Paper WP/13/42. Krishna, P., J. Poole and M. Senses (2014). Wage Effects of Trade Reform with Endogenous

Worker Mobility, Journal of International Economics, 93 (2), 239–252. Launov, A and K. Wälde, (2016), The Employment Effect of Reforming a Public Employment

Agency, European Economic Review, 84, 140–164. Lehmann, H and J. Kluve, (2010). Assessing Active Labour Market Policies in Transition

Economies, AIEL Series in Labour Economics,in: Floro Ernesto Caroleo & Francesco Pastore (ed.). The Labour Market Impact of the EU Enlargement. A New Regional Geography of Europe?, edition 1, chapter 11, pages 275-307.

http://ec.europa.eu/social/BlobServlet?docId=17989&langId=enhttp://ec.europa.eu/social/BlobServlet?docId=17989&langId=enhttp://www.euklems.net/

25

Melitz, M.J. (2003). The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity. Econometrica, 71, 1695-1725.

Merz, M. (1995). Search in the Labor Market and the Real Business Cycle, Journal of Monetary Economics, 36, 269–300.

Moore, M. P., and P. Ranjan (2005). Globalisation vs Skill-Biased Technological Change: Implications for Unemployment and Wage Inequality, The Economic Journal, 115, 391–422

Mosseri-Marlio, W. (2016). The Future of Public Services: Digital Jobcentres, Reform, July. OECD (2018), Employment by Education Level (Indicator). OECD (2016), Education at a Glance 2016: OECD Indicators, OECD Publishing, Paris. OECD (2015). Strengthening Public Employment Services, Paper prepared for the G20

Employment Working Group, Istanbul, Turkey. ONS (2016). Analysis of the UK Labour Market – Estimates of Skills Mismatch Using

Measures of Over and Under Education: 2015. Petrongolo, B. and C.A Pissarides (2001). Looking into the Black Box: A Survey of the

Matching Function, Journal of Economic Literature, 39, 390–431. Riley, R., H., Bewley, S., Kirby, A., Rincon-Aznar and A., George (2011). The Introduction

of Jobcentre Plus: An Evaluation of Labour Market Impacts, Department for Work and Pensions, Research Report No 781.

Schlotter, M (2008), Origins and Consequences of Changes in Labour Market Skill Needs, European Expert Network on Economics of Education, Analytical Report No. 4.

The World Bank (2016). Doing Business: Measuring Business Regulations. Vallanti, G. (2015). International Capital Mobility and Unemployment Dynamics: Empirical

Evidence from OECD Countries, LLEE Working Paper Series No. 123. van Vliet, O. and K. Caminada (2012), Unemployment Replacement Rates Dataset Among 34

Welfare States 1971-2009: An Update, Extension and Modification of Scruggs, Welfare State Entitlements Data Set. NEUJOBS Special Report No. 2.

WDI – The World Bank (2016). World Development Indicators.

http://doi:10.1787/26f676c7-enhttp://dx.doi.org/10.187/eag-2016-enhttps://www.ons.gov.uk/releases/%20analysisoftheuklabourmarketestimatesofskillsmismatchusingmeasuresofoverandundereducation2015https://www.ons.gov.uk/releases/%20analysisoftheuklabourmarketestimatesofskillsmismatchusingmeasuresofoverandundereducation2015http://www.doingbusiness.org/datahttps://data.worldbank.org/%20indicator/NE.TRD.GNFS.ZS?locations=US

Appendix

Table 1. Steady State Solution Values Showing Effects of Policies

Variables Bench-mark

raised by 20%

b raised by 60%

f raised by 43%

jc lowered by 50%

FRP

GDP 1.000000 0.931742 1.002466 0.999790 1.036438 1.037713

Price Level P 0.992875 1.031794 0.992477 0.992909 0.987099 0.986901

Domestic Absorption Y 1.000000 0.947660 1.001891 0.999839 1.027940 1.028918

Foreign Absorption (Net Exports) 0.000000 -0.015919 0.000575 -0.000049 0.008498 0.008795

Aggregate Unemployment S 0.060000 0.063130 0.072914 0.060005 0.040771 0.049570

Unskilled Unemployment ulS 0.035000 0.036867 0.043819 0.035000 0.024412 0.030156

Skilled Unemployment shS 0.010458 0.011229 0.011657 0.010463 0.004328 0.005191

Mismatched Unemployment slS 0.014542 0.015033 0.017438 0.014542 0.012031 0.014223

Aggregate Dispersion Index ASDI 0.148235 0.148749 0.148786 0.148069 0.144547 0.144755

Mismatched Ratio MMR 0.260134 0.260435 0.261835 0.260085 0.258674 0.259447

Aggregate Employment N 0.730000 0.725262 0.732317 0.729762 0.757796 0.759003

Unskilled Employment ulN 0.367182 0.365214 0.370408 0.367009 0.379211 0.381608

Skilled Employment shN 0.267094 0.265174 0.267740 0.267068 0.278662 0.279026

Mismatched Employment slN 0.095724 0.094873 0.094169 0.095685 0.099923 0.098368

Aggregate Participation X 0.790000 0.788392 0.805231 0.789767 0.798566 0.808572

Unskilled Participation ulX 0.402182 0.402081 0.414227 0.402009 0.403623 0.411764

Skilled Participation shX 0.277552 0.276404 0.279397 0.277531 0.282990 0.284218

Mismatched Participation slX 0.110266 0.109907 0.111607 0.110227 0.111954 0.112591

Low-Tech Job Matches l

0.010395 0.010283 0.010356 0.010390 0.011322 0.011232

High-Tech Job Matchesh

0.002404 0.002387 0.002410 0.002404 0.002508 0.002511

On the Job Search Success Rate 0.001137 0.001081 0.001065 0.001136 0.001739 0.001632

Unskilled Bargained Wage ulw 0.395538 0.370604 0.401990 0.395297 0.404243 0.408645

Skilled Bargained Wage shw 1.140605 1.067593 1.147088 1.140322 1.168087 1.172421

Mismatched Bargained Wage slw 0.408189 0.382521 0.415929 0.407962 0.418086 0.424204

Market Tightness, Low-Tech jobs l 0.680934 0.607149 0.442090 0.680272 1.492902 0.990756

Market Tightness, High-Tech jobs h 0.212552 0.195697 0.192716 0.212486 0.456710 0.415057

Low-Tech Labour lH 0.028581 0.028407 0.028674 0.028568 0.029583 0.029625

High-Tech Labour hH 0.029999 0.029785 0.030071 0.029996 0.031292 0.031333

Unskilled Job Match Value ul

1.016680 0.958417 0.788662 1.016100 0.741584 0.578388

Skilled Job Match Value sh

7.104328 6.816841 6.764720 7.103233 5.206919 4.963804

Mismatched Job Match Value sl

1.354521 1.287219 1.176601 1.354066 0.997617 0.873552

Domestically Soled Variety Quantity dy 13.606591 15.770846 13.642838 13.603496 14.142707 14.161482

Exported Variety Quantity xy 5.831396 2.640986 5.846420 5.830113 6.053250 6.061006

Imported Variety Quantity *y 0.212680 0.105489 0.212699 0.212679 0.212958 0.212968

Domestic Price of Variety dp 1.099391 1.092488 1.098762 1.099444 1.090306 1.089996

Welfare (Household Utility) -51.54553 -57.91482 -51.79542 -51.55107 -48.22559 -48.39513

Table 2. Pre-Reform vs. Post-Reform De-Globalisation

Pre-Reform solution at = 1 (benchmark) (with: I g = 0.003, = 3.800, = 0.00) Pre-Reform solution as rises (with: I g = 0.003, = 3.800, = 0.00) Post-FRP solution as rises (with: I g = 0.003, = 3.800, = 0.00) Post-FRP solution as rises (with: I g = 0.006, = 3.800, = 0.00) Post-FRP solution as rises (with: I g = 0.003, = 3.825, = 0.00) Post-FRP solution as rises (with: I g = 0.003, = 3.800, = 0.25)

Table 2. Pre-Reform vs. Post-Reform De-Globalisation (continued)

WSF WP GlobLabWS #2_2018_Molana Montagna_Onwordi_CoverWSF WP GlobLabWS #2_2018_Molana Montagna_OnwordiMismatch and Firm Heterogeneity Paper(CM-Draft 8- 10_12_2018)1. Introduction2. The Model2.1. Households

2.2. Vacancies and matching2.2.1. Low-tech job agency

2.4. The final good sector2.5. The intermediate good sector3. Calibration4. De-globalisation5. Reforming the Labour Market6. Post-Reform Trade De-Globalisation7. Summary and Conclusions

Appendix_Tables_27-11-18