View of The Effect of Over-education and Overskilling on Economic Growth across States in Malaysia: An Empirical Evidence

(1)

The Effect of Over-education and Overskilling on Economic Growth across States in Malaysia:

An Empirical Evidence

Zainizam Zakariya*1_{, Normala Zulkifli}2_{, Khoo Yin Yin}3_{, AlifNawi}4

*1,2,3_{PhD, Faculty of Management and Economics, Universiti Pendidikan Sultan Idris (UPSI), 35900 Tanjung}

Malim,Perak

4_{PhD, School of Education, Universiti Utara Malaysia (UUM), Sintok, 06010 Bukit Kayu Hitam, Kedah}

Article History: Received: 10 November 2020; Revised: 12 January 2021; Accepted: 27January 2021;

Published online: 05April 2021

Abstract: This paper explores the incidence and the outcome of educational and skill mismatchoneconomic growth across

state in Malaysia from2006 and 2012.The mismatch indicators were gauged using the Job Analysis (JA) and the mode method.Using micro cross-section data from Labour Force Survey (LFS) between 2006 and 2012, overqualification (underqualification)and overskilling (underskilling) were reported between13 (20) and 19 (34) percent.Results for Fixed Effect (FE) regression demonstrated overqualification and overskilling had a favourable impact on regional growth. By contrast, the growth was negatively associated with an increase in undereducation and underskilling incidence. The findings depict that the economic performance at the regional level in Malaysia is associated with an increase in overqualification and overskilling. The presence of such incidence, therefore, may not be a sign of inefficient public investment and resources allocated to education are in fact economically beneficial at a macro-level.

Keywords:over qualification, over skilling, economic growth, state, Malaysia

Introduction

Researchers in traditional labour market have argued that workers seek for jobs on the large or regional rather than small market due greater job opportunities provided in the former than in the latter. Yet, due to spatial mobility constraints experienced by some workers, they have looked for work on the local or small labour market(Hensen, de Vries, & Cörvers, 2009; Cabus & Somers, 2018). Job seekers whoexperienced flexibilitylimitation tend to end up in a local job that below than their actual education or skill background, resulting in overqualification(Büchel & Battu, 2002; Kulkarni, Lengnick-Hall&Martinez, 2015) or overskilling (Zakariya, Abdul, et al., 2017; Zakariya & Yin, 2017).

Assessments of the degree of theconsequence of overqualification and overskillingbetween large and small labour marketmay seemdecisive for policymakers as such incidences are typically ending up in negative rather than positive outcomeseither at an individual nor at a firm level.1_{The negative outcome at both levels may drive}

down local economic performance and leads to an unequal distribution of economic development between smaller and larger region. Thismay seem to be truly in the context of Malaysia as differences in educational attainment, unemployment, occupationor education and skills utilisationhas led to regional income inequalities and unbalanced growth among states(Yussof & Kasim, 2003; Ragayah, 2008; Saari, Dietzenbacher, & Los, 2014; Abdullah, Doucouliagos, & Manning, 2014; Hutchinson, 2017;Zakariya, Hermanssons, Yin, Fazlin, & Noor, 2019).

Nevertheless, if the supply of highly educated workers is not in line by demand at the state labour market level, then the impact of education and skill on state Growth Domestic Product (GDP) may not as high as expected relative to if the state werefully utilised the education and skills of all the mismatched workers(McGowan and Andrews, 2017; Adrian, Desislava, Ganev, & Aleksiev, 2018).2_{Yet, educational and}

skill mismatch incidencesmayalso in turn lead to appositive outcome on state economicgrowth. This is because the overqualified and overskilledare typically have accumulatedmore schooling and skill, hence, more

1_{At an individual-level study, both overeducated and overskilled earn significantly lower than their comparable}

well-matched(see review in Leuven & Oosterbeek, 2011) and greater job dissatisfaction (Fleming & Kler, 2008; Di Paolo & Mañé, 2016; Verhaest & Verhofstadt, 2016). Some studies found over-education at the workplace improves firm level productivity (Jones, Jones, Latreille, & Sloane, 2009; Kampelmann & Rycx, 2012; Mahy, Rycx, & Vermeylen, 2015;Philipp Grunau, 2016).

2_{A study from Wald (2004) Canada showed that over-qualification resulted in approximately 2 percent or $20}

billion reduction in the Canadian Gross Domestic Product (GDP) due to lower tax revenues among the overqualified workers.

(2)

Zainizam Zakariya*, Normala Zulkifli, Khoo Yin Yin, AlifNawi

productive than that of the well-matched group(Sloane, Battu, & Seaman, 1996; Chiswick & Miller, 2010; Sánchez-Sánchez, McGuinness, 2015). Consequently, they may have driven up local economic growth.Therefore,any economic impact of overqualification and overskillingat the regional levelcould be possible. As far as we are concerned, there has no study examined the link between educational-skillmismatch and economic growth. Instead, there has very few study examined the linkages between educational mismatch on growth (Ramos, Surinach, & Artís, 2012; Zakariya, Hermanssons, Yin, Fazlin, & Noor,2019). It should be acknowledged that both educational and skill mismatch are two difference phenomenon in the labour market as each incidence captures different dimension (Mavromaras, Mcguinness, O’Leary, Sloane, & Fok, 2010; Zakariya, Abdul Jalil, & Yin Yin, 2017; Zakariya & Yin, 2017).

Therefore, this paper aims to explore the incidence and the consequence of not only aggregate overqualificationbut also aggregate overskilling on economic growth across states in Malaysia between 2006 and 2012. If overskilling reduces workers’ own productivity, this negative outcomemight be translated into decreasing output at macro-level, i.e. GDP growth, especially state with higher proportion of overskilling in the labour market. In doing so, the rest of the paper is structured as follows. Section two reviews past studies related to the outcomes ofoverqualificationand overskilling at the aggregate level if any.Dataset, measurement and empirical method are outlined in section three.Sectionfourfocuses on empirical findings whereas section five highlightsdiscussion and conclusion of the study.

Overqualification, Overskillingand Regional Economic Growth

The typical findings ofnegative outcomes of overqualification incidence on workers tend to have an adverse effect at a firm or a country-level. At the firm level, for example, Tsang (1987)revealed thatover-education indirectly reduced firm productivity in Bell companies thru job dissatisfaction mechanism. A one-year increase in surplus education led to a losing in 8.4 percent firm output. In Germany, Philipp (2016)demonstrated that under-education rather than overeducation impair firm productivity. Other studies demonstrated that overqualification decreased workplace average pay (Belfield, 2010) and led to workplace dis-harmonization in terms of absenteeism and quit rate (Jones, Jones, Latreille and Sloane, 2009; Belfield, 2010).

There are few studies, however, found that overqualification results in workplace improvement with respect to financial performance (Jones et al., 2009) and firm productivity (Kampelmann & Rycx, 2012; Mahy et al., 2015). For instance, Mahy et al. (2015)found that higher incidence of over-educationat workplaces tend to raise firm productivity, especially at firms with a greaterproportion of skilled workers at the workplace in high-tech/knowledge-intensive industries. Thesemay be due to the overeducated have more skills and greater educational attainment than their comparable well-matched, hence more productive (Hartog, 1988; Sloane et al., 1996; Hartog, 2000). Thismight have an impact on other workers’ productivities at the workplace, hence, improving establishment-level productivity. Battu et al. (2003) and Mohamed Noor et al. (2017) for example found workers who employed with co-workers who have more schooling than theirs boost own earnings. Indeed, few studies found firms in regions with greater proportion of human capital stocks are more productive than firms in regions with less one (Acemoglu & Angrist, 2001; Moretti, 2004; Liu, 2007; Sand, 2013; Mohamed Noor et al., 2017).

Up to a certain point, negative outcomes of over-education at the firm level may possibly reduce national income than would be the case if all the skills and knowledge of the overqualified or overskilled workers were fully exploited within the economy. Unfortunately, up to our knowledge, there is very little study available at a macro-level. McGowan and Andrews (2017)found overskilling decreases labour productivity across 19 OECD countries. Guironnet and Peypoch (2007) and Adrian et al., (2018) in their respective studies in France and European countries found skill mismatch tends to reduce a country’s aggregate productivity.

With respect to the outcomes of over-education on growth, there seems unconvincing findings. A study byJaoul-Grammare and Guironnet (2009)provides a limited evidence of negative causality of over-education on France’s economic growth in the short run. Specifically, the study revealed that higherproportion of overeducated workers without degree in the workforce was negatively associated with growth in France.3

Instead, Ramos et al. (2012) demonstrated thatall educational mismatch indicators, i.e. - education, over-qualification and years of over-schooling are positively associated with growth, ranges from 3 to 13 percent across nine European countries. Of the three indicators, the effect was greater for the overeducation incidence

3_{By contrast, the authors found no evidence of causality relationship for the three different groups of}

overeducation - the share of overeducated workers of the higher education (SOHE), overeducated workers of the Higher Education (OHE), and overeducated workers without any degree of higher education (OWHE)

(3)

(model with country and time fixed effect).In a recent study, Zakariya et al. (2019)examined the outcome of aggregate overqualification on growth across region in Malaysia from 2005 to 2017 using Dynamic Panel Data (DPD) time series analysis. In the study, the aggregate overqualification was measured as the fraction of workers with at least a bachelor’s degree qualification who were employed in an occupation below than the professional job level. The authors found strong evidence of negative outcome of the aggregate overqualification on growth across region in Malaysia although the magnitudes of the effect were smaller (between 0.02 and 0.03).

Perhaps, theinconclusive results maybepartly explained by differences in the measurement, method and dataset employed. The mismatch could have different outcomes in developing economies due to lower income, but education levels are rising faster than the growth as noted earlier. Therefore, they might not reap as much benefit from higher education investment as they might hope for due to slower job creations leave the underutilisation of the highly educated person. Nevertheless, to some extent, the negative consequence of over-education on growth may depict there would be growth penalty for not being fully utilised the knowledge and skills of highly educated workers at the regional labour market.

Dataset, Measurement and Methodology of the Study

Dataset and Measurements of Aggregate Educational and Skill Mismatch

To measure educational and skill mismatch at aggregate level, we employed data from Labour Force Survey (LFS), 2006 – 2012provided by Department of Statistics Malaysia (DoSM). The advantage of the LFS’s sample lies in the fact that a number of key variables of employed persons such as educational and qualification level, occupations and types of industry are recorded using a homogeneous classification, allowing us to calculate the rate of educational and skill mismatch and other variables in a comparable way across 12 states and 2 federal territories (Kuala Lumpur and Labuan).4_{This allows us to measure the rate of overqualification}

(underqualification) and overskilling (underskilling) using traditional method of the Job Analysis (JA) and the mode method.5_{In particular, we calculate first both the educational and skill mismatch at the individual level,}

andsubsequently aggregate them into regional level to obtain regional indicators of the incidence following Ramos et al. (2012). Meanwhile, macro level data, i.e. - GDP per capita (Y) and capital (K) across state, each was extracted from the DoSMand the Malaysia Investment Development Authority (MIDA). Both variables were measured in logarithm form based on 2010 constant price (Ringgit Malaysia).6

Table 1.Descriptive statistics of the key variables

Year LFS 2006 (state = 15, N = 28,783) LFS 2009 (state = 15, N = 47,292) LFS 2012 (state = 15, N = 47,319) Total

Mean SD Mean SD Mean SD Mean SD

lncgdp (lnY) 10.22261 1.08003 10.32632 1.11174 10.51957 1.09245 10.35617 1.07686 lnccap (lnK) 21.13621 1.99464 20.06849 2.80135 20.46602 2.46939 20.55690 2.42933 lnlab (lnL) 13.06752 1.08024 13.13627 1.05923 13.26864 1.08894 13.15748 1.05485 Schooling (sch) 10.93661 0.60788 10.98208 0.89124 10.96479 0.79628 10.96011 0.89145 aggoverq_ja (%) 16.26282 3.16608 16.31288 2.54843 16.49543 2.84571 16.35704 2.80050 aggunderq_ja (%) 17.26322 2.13686 20.61598 4.25371 22.95665 4.92023 20.27862 4.52729 aggoverq_mode (%) 16.21273 2.13742 18.73011 3.18705 20.31717 2.51860 18.42001 3.10235

4_{Under the LFS, educational level and highest qualification attained are classified following the 1997}

International Standard Classification of Education (ISCED) while occupations are classified in accordance with the 2008 International Standard Classification of Occupation (ISCO).

5_{Details of the mode method are outlined inRamos et al. (2012).}

6_{Due to data on capital formation was not available across state, the variable K was measured using capital}

(4)

aggunderq_mode

(%) 28.37468 4.01411 35.06449 4.33419 36.79993 4.50689 33.41303 5.57326 aggoversk (%) 11.77063 2.22256 14.17598 2.13673 15.03757 2.17938 13.66139 2.54762 undersk (%) 20.75513 3.67920 24.74185 5.09061 29.08273 4.96194 24.85990 5.67578

Table 1presents the mean and standard deviations of some key variables used in the paper. On average, GDP per capita (lnY) is 10.36 (s.d: 1.07), capital per capita (lnK) and labour (lnL)per year is 20.56 (s.d – 2.43) and 13.16(s.d:1.05) per year, respectively. With respect to human capital endowment, approximately workers have completed nearly 11 years (s.d: 0.89 year) of schooling.7_{This was equivalent to SPM qualification.Turning to}

educational mismatch, in general, the proportion of aggregate overqualificationwas a bit higher for the mode (aggoverq_mode) than the JA method (aggoverq_ja) with the corresponding figure of 16% against 18% each. By contrast, the incidence of aggregate underqualificationwas a much lower for the latter (aggunderq_ja), roughly 20% relative to 33% for the former method (aggunderq_mode). With respect to skill mismatch, aggregate overskilling (aggversq) and underskilling (aggundersq) represented about 13.7% and 25% of the total employed person.

Looking first at the JA method, Figure 1 shows in general, the proportion of overqualification does not have a clear pattern. In some state, overqualification remained higher between 2006 and 2002 as illustrated in Perlis, Terengganu and Kuala Lumpur. Other states demonstrated a decline trend in Kedah, P. Pinang, N. Sembilan and Perak whereas the incidence remained lower and stable in Sarawak, Selangor, Sabah and Johor.

Figure 1. The Proportion of Over and Underqualification under the JA Approach across State (%)

Turning to mode method (Figure 2), there seems a clear trend where the incidence of overqualification showed an increase trend between 2006 and 2009 in almost states and the incidence remained stable after cross all states (albeit for Kuala Lumpur and Labuan.

7_{Following the 1997 ISCED, we converted levels of qualification among the employed persons (for example,}

UPSR, PMR/SRP and SPM) into number years of schooling completed. For instance, UPSR requires 6 years of education while SPM requires 11 years of education.

15 20 25 30 15.5 16 16.5 17 17.5 12 14 16 18 20 14 16 18 20 22 14 16 18 20 22 15 20 25 30 14 16 18 20 22 10 15 20 25 14 16 18 20 22 18 20 22 24 14 16 18 20 10 15 20 25 30 10 15 20 25 30 16 18 20 22 18 20 22 24 26 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012

Johor Kedah Kelantan Melaka

N. Sembilan Pahang P. Pinang Perak

Perlis Selangor Terengganu Sabah

Sarawak WPKL WP Labuan

peroe_om perue_om

Year

Source: Authors' own calculation

The proportion of over and underqualification under the JA approach across state (%) Figure 1 P er ce nt (% )

(5)

Figure 2.The Proportion of Over and Under qualification under the Mode Method Across State (%)

With respect to underqualification, such incidence always outnumbered overqualification in almost all states regardless of measurements. In fact, Figure 2 shows underqualification was always greater than overqualification across all states over the period of 2006 – 2012.

Nevertheless, the proportion of overskilling demonstrates a steady increase across almost states (albeit for WP Kuala Lumpur) between 2006 and 2012. Instead, the incidence of underskilling exhibits a decline trend over the same period for all states.

Figure 3. The Proportion of Over and UnderskillingAcross State (%)

15 20 25 30 35 20 25 30 35 20 25 30 35 15 20 25 30 35 10 20 30 40 15 20 25 30 35 10 20 30 40 10 20 30 40 15 20 25 30 35 15 20 25 30 35 15 20 25 30 35 20 25 30 35 40 10 20 30 40 15 20 25 30 10 20 30 40 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012

peroe_mode perue_mode

Year

The proportion of over and underqualification under the mode method across state (%) Figure 2 P e rc e n t (% ) 10 15 20 25 30 10 15 20 25 30 15 20 25 30 10 15 20 25 10 15 20 25 30 10 20 30 40 10 15 20 25 10 20 30 40 15 20 25 30 10 15 20 25 10 15 20 25 30 10 20 30 40 10 20 30 40 12 14 16 18 15 20 25 30 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012

Agg overskillling Aggregate underskilling

Year

The proportion of over and underskilling across state (%) Figure 3 P er ce nt (% )

(6)

Empirical Framework

Due to the nature of data used here, this paper employs fixed effect (FE) regression to ascertain the effect of educational and skill mismatch on growth. This is because the method permits us to control for unobservable heterogeneity through the inclusion of state and time-fixed effects. In particular, GDP per capita (𝑙𝑛𝑌) between 2006 and 2009 and between 2006 and 2009 is regressed on the initial level of GDP per capita (𝑙𝑛𝑌𝑖,𝑡−𝜏) and the

educational-skill mismatch indicators for the different sets of regions both periods. Following Ramos et al., (2012), the model can be written as below:

(𝑙𝑛𝑌𝑖,𝑡− 𝑙𝑛𝑌𝑖,𝑡−𝜏)/𝑥 = 𝛼 + 𝛽 ∙ 𝑙𝑛𝑌𝑖,𝑡−𝜏+ 𝜕 ∙ 𝑙𝑛𝐾𝑖,𝑡−𝜏+ 𝛿 ∙ 𝑙𝑛𝑠𝑐ℎ𝑖,𝑡−𝜏+ 𝛾 ∙ 𝑥𝑖,𝑡−𝜏+ 𝜂𝑡+ 𝑣𝑖(1)

where𝑣𝑖= 𝜇𝑖+ 𝜀𝑖,𝑡

where 𝑙𝑛𝑔𝑑𝑝𝑖𝑡 is the natural log ofreal GDP per capitain regioni at yeart;𝑙𝑛𝐾𝑖and 𝑙𝑛𝑠𝑐ℎ𝑖 represent the natural

log of real capital per capita and years of schooling of the working population in region i, respectively; 𝑥𝑖,𝑡−𝜏denotes educational-skill mismatch indicators in state i at year t;𝜂𝑡is a time-fixed effect; 𝜇𝑖a region fixed

effect; and 𝜀𝑖,𝑡is a random error term that varies across region and time periods.

Empirical Findings

Table 2 presents the results of Fixed Effect (FE) estimator across models with the different explanatory variables. In models (1), GDP per capita was regressed on initial GDP per capita and traditional growth model, i.e.-𝑙𝑛𝐾and 𝑙𝑛𝑆. Indicators ofeducational and skill mismatch among the working population are included in models (2) to (4). Specifically,percentage of overqualification following objective method (𝑙𝑛𝑜𝑒_𝑜𝑚)and mode method (𝑙𝑛𝑜𝑒_𝑚𝑚) is included in model (2) and model (3), respectively while in model (4), we controlled for percentage of overskilling (𝑙𝑛𝑝𝑒𝑟_𝑜𝑠). The significant of F-test across models indicating that the FE estimator is preferable than the Ordinary Least Square (OLS) estimator.

Table 2.Growth effect of overqualification and overskilling with regional, industry and time fixed-effects

GDP per capita (𝒍𝒏𝒀) Model 1 Model 2 Model 3 Model 4

Initial GDP per capita

(𝑙𝑛𝑌𝑡−𝜏) -0.2167 *** -0.2081 *** -0.2163 *** -0.2078 *** (0.0565) (0.0593) (0.0579) (0.0587) lncapital (𝑙𝑛𝐾) 0.0122 *** 0.0105 *** 0.0125 *** 0.0108 *** (0.0015) (0.0014) (0.0014) (0.0015) Years of schooling (𝑙𝑛𝑆) 0.2012 *** 0.1228 *** 0.1698 *** 0.1977 *** (0.0645) (0.0586) (0.0773) (0.0632) % overqualified workers (𝑙𝑛𝑜𝑒_𝑜𝑚) 0.0610 *** (0.0127) % overqualified workers (𝑙𝑛𝑜𝑒_𝑚𝑚) 0.0180 *** (0.0036) % overskilled workers (𝑙𝑛𝑝𝑒𝑟_𝑜𝑠) 0.0500 *** (0.0112) N 45 45 45 45 Number of groups 15 15 15 15 Adjusted R-square 0.9961 0.9961 0.9964 0.9961 Rho () 0.955 0.956 0.963 0.956 F-test (all ui=0) 24.37*** 23.10*** 23.20*** 23.39***

(7)

Results from the Table 2 reveals some interesting findings. The coefficient of initial GDP per capita (𝑙𝑛𝑌𝑖,𝑡−𝜏)

is negative and significant at 0.01 across all models, indicating that a process of regional convergence has occurred during the period under review. This process is still apparent when additional covariates are controlled for together. The effect of 𝑙𝑛𝐾 on growth is positive and statistically significantly different from zero at 0.01. Yet, the magnitude is very small, i.e. – less than 0.02. Likewise, the traditional indicator of human capital, i.e. 𝑙𝑛𝑆 across models is always positive and has a strong impact on economic growth. Other factors being held constant, an increase of additional one year of schooling completed among the working population leads to a raise in GDP per capita by approximately between 0.13 (e0.1228_{) and 0.22 (e}0.2012_{) percent.}8

Focus on our main variable interest, model (2) to model (4) show that the coefficients of overqualification and overskilling are positive and statistically significantly different from zero at the 1% level indicating that both incidences have a strong impact on regional economic growth. The magnitudes of the effects are however somewhat depending on measurement used. Using the objective method, the coefficient of 𝑙𝑛𝑜𝑒_𝑜𝑚(model 2) is 0.0610, suggesting that one percent increase in the percentage of overqualified workerslead to an increase of approximately 0.06 percent in regional GDP per capita. Using the mode method (model 3), the coefficient of 𝑙𝑛𝑜𝑒_𝑚𝑚 is 0.0180. Perhaps, lower return may reflect higher incidence of overqualification produced by the mode method as compared to the objective method. In model (4), the coefficient of𝑙𝑛𝑝𝑒𝑟_𝑜𝑠 is positive and statistically significant at the 1% level. One percent increase in the percentage number of overskilling at the regional level will lead to an increase of about 0.05 percent in GDP per capita.

Another point that emerging from Table 2 is that the impact of schooling on growth tends to decline once overqualification is included in the regression. For example, when 𝑙𝑛𝑜𝑒_𝑜𝑚 and 𝑙𝑛𝑜𝑒_𝑚𝑚 is included together, respectively in the Model 2 and Model 3, the coefficient of 𝑙𝑛𝑆 is about 8 and 5 percentage points lower than the previous model. Two additional tests have been carried out in order to test the robustness of the results to changes in the econometric specification. First, we carried out the log likelihoods ratio test from both models and found that adding 𝑙𝑛𝑜𝑒_𝑜𝑚 in model 2 and 𝑙𝑛𝑜𝑒_𝑚𝑚 in model 3 as a predictor variable result in a statistically significant improvement in model fit.

We also perform parameter test of schooling and overqualification and reject the null hypothesis that the coefficients for both parameters are jointly equal to zero. All these may indicate that overqualification effects pick up some of effects of schooling on GDP per capita.This may not be surprisingly due to the way overqualification is measured, i.e. - based mainly on years of education for any given occupation.By contrast, when overskilling is controlled for as (model 4), the growth impact of schooling does though remain similar to model 1.

In Table 3, we replace both overqualification and overskilling at the regional level with underqualification and underskilling. Threespecificationsare examined. In model 5, the underqualification based on the objective method (𝑙𝑛𝑢𝑒_𝑜𝑚) is added together with the basic model (Model 1) whereas in Model 6, underqualification is represented by the mode method (𝑙𝑛𝑢𝑒_𝑚𝑚). In Model 7, underskilling(𝑙𝑛𝑝𝑒𝑟_𝑢𝑠)is replaced for underqualification.The results with respect to 𝑙𝑛𝑌𝑖,𝑡−𝜏, 𝑙𝑛𝐾 and 𝑙𝑛𝑆 seem similar to Model 1 in terms of the sign

and significant level with one exception where the coefficient of 𝑙𝑛𝑆 in Model 6 has now turned out to be negative but insignificant in both models. Therefore, the conclusions almost remain unchanged

.

8_{Since the lnSis in the natural logarithmic form, the percentage point effect (PE) is obtained using the following}

formula:

PE = (eβ – 1) x 100, where β is the coefficient estimate.

The percentage point effect will be used throughout the discussion in this chapter. Instead, the coefficients of other variables represent the elasticity values.

(8)

Table 3.Growth effect of underqualification and underskilling with regional, industry and time fixed-effects

GDP per capita (𝒍𝒏𝒀) Model 5 Model 6 Model 7

Initial GDP per capita (𝑙𝑛𝑌𝑡−𝜏) -0.2138 *** -0.2367 *** -0.2085 ***

(0.0549) (0.0560) (0.0510)

lncapital (𝑙𝑛𝐾) 0.0106 ** 0.0155 * 0.0100 *

(0.0052) (0.0090) (0.057)

Years of schooling (𝑙𝑛𝑆) 0.2054 *** -0.3634 0.1476 **

(0.0614) (0.2230) (0.0688)

Percentage of underqualified workers (𝑙𝑛𝑢𝑒_𝑜𝑚) -0.0819 *** (0.0213)

Percentage of underqualified workers (𝑙𝑛𝑢𝑒_𝑚𝑚) -0.2772 **

(0.1211)

Percentage of underskilled workers (𝑙𝑛𝑝𝑒𝑟_𝑢𝑠) -0.1576 ***

(0.0462) Constant 11.5950 *** 13.6998 *** 11.9833 *** (1.4706) (1.5221) (0.0142) N 45 45 45 Number of groups 15 15 15 Adjusted R-square 0.9964 0.9970 0.9964 Rho () 0.996 0.9969 0.9957

Note: *, ** and *** significant at 0.1, 0.05 and 0.01 respectively Robust standard error in parenthesis

Turning now on the effect of underqualification, the coefficient of 𝑙𝑛𝑢𝑒_𝑜𝑚 and𝑙𝑛𝑢𝑒_𝑚𝑚respectively for Model 5 and 6 are negative and significantly with at least at the 5% level. This means that underqualification has unfavourable impact on growth. Yet, the impact is 3 times lower for the objective (Model 5) than for the mode measure (Model 6). Other factors being equal, one percentage point increase in aggregate underqualification, GDP per capita at the regional level will decline by around 0.08 percent for the former and about around 0.24 percent for the latter. Indeed, the growth impact of schooling has turned out to be negative, but not statistically significant. When underqualification is replaced with underskilling, Model 7 demonstrates that such variable (𝑙𝑛𝑝𝑒𝑟_𝑢𝑠) has also negatively associated with growth over the period of study. The GDP per capita at the regional level will be decreased by approximately 0.15 percent, all things equal.

Discussion and Conclusion

This study explored the incidence and the outcome ofaggregate educational and skill mismatch among employed personsacross region in Malaysia between 2006 and 2012.The study employs conventional methods of the objective and mode method in calculating the aggregate mismatch by using micro survey data, i.e – LFS and then aggregated into regional level. Between 13and 18% (21 and 33%)of employed workers were deemed overqualified (underqualified)whereas around 14% were considered as being over skilled workers. Both incidences werecomparable higher in K. Lumpur and Selangor than other states.

The findings somewhat contrast to our expectation as the developed state could experience a lower rate of aggregate mismatch than the less developed state due to the former rather than the latter can provide more suitable jobs for the highly educated person.Perhaps, the findings reflect larger numbers of vacancies in the developed states areoffset by a larger number of job searchers (Mcgoldrick & Robst, 1996). This seems to be true as these states have as many as higher educational institutions relative to other states (Ministry of Higher Education Malaysia, 2018), therefore provide more highly educated job seekers.

After a range of statistical tests performed, we employed fixed effect panel data approach to investigate the growth outcome of educational and skill mismatch at the regional level. Two specifications were examined. In the first specification, we controlled for overqualification and overskilling as shown in Model 2 – Model 4.Regardless of any model, there was strong evidence of the positive impact of both incidenceson regional growth.In specification 2, we replaced underqualification and underskilling for overqualification and overskilling, respectively. The results both incidences were negatively associated with regional economic

(9)

growth. As such, our results with respect to over- and underqualification seem in line with finding from Ramos et al. (2012). The magnitudes of the impact were much higher for the mode than the objective method. This may reflect the mode tend to generate more (less) overqualified (underqualified) than the objective method.

Nevertheless, the positive impact of overqualification and overskilling may be partly due to the stylized fact that the overqualified and overskilled workers tend to earn a wage premium compared to their co-workers who have less schooling but in well-matched jobs (Sicherman & Galor, 1991).9_{They might still more productive than}

the latter and this might have an impact on other workers’ effort. Battu et al. (2003) and Mohamed Noor et al. (2017) for example find that workers who employed with co-workers who have more schooling than theirs boost own earnings. Otherwise, the overqualified workers tend to have accumulated more skills and greater educational attainment than their comparable well-matched, hence more productive (Hartog, 1988; Sloane et al., 1996; Hartog, 2000). As mentioned in human capital externality hypothesis, firms and regions with greater human capital stock tend to be more productive than firms and regions with less human capital stocks accumulation (Acemoglu & Angrist, 2001; Moretti, 2004; Liu, 2007; Sand, 2013; Mohamed Noor et al., 2017). This spillovers can be translated into improving establishment-level productivity, hence, contagious at macroeconomic level.

Moreover, the fact that the mismatched workers have accumulated more skill and schooling may suggest that the human capital of mismatched workers may contribute to public benefits associated with higher levels of the national human capital stock (Hartog, 1988; Sloane, Battu, & Seaman, 1996; Chiswick & Miller, 2010). A country with a high stock of human capital tends to exhibit higher labour productivity (Mankiw, Romer, & Weil, 1992; Hanushek & Wößmann, 2010; Breton, 2011; Hanushek, 2013), be more innovative (Lucas, 1988; Romer, 1990, 1994)and be better at adopting new technologies (Benhabib & Spiegel, 1994).

To some extent, the positive impacts of overqualification and overskilling on growth may not reflect a waste of investment in higher education in Malaysia. This is becausethe coefficient of schoolingis always positively associated with growth regardless of the model specification even after controlling for both variables. Moreover, it is an exaggeration to say that the region with greater overqualified or overskilled worker may indicate higher levels of human capital accumulation, hence, higher labour productivity(Hanushek, Jamison, Jamison, & Woessman, 2008; Breton, 2011; Hanushek & Wößmann, 2010; Hanushek, 2013), increase the innovative capacity of the economy (Lucas, 1988; Romer, 1990, 1994) and transmission of knowledge and new technologies (Nelson & Phelps, 1966; Benhabib & Spiegel, 1994; Hanushek et al., 2015) than countries with lower levels of human capital stock.

Moreover, the findings from this paper may suggest that the growth may no longer a function solely of the supply side (educational attainment of workers) as done in many previous studies (Yussof & Zakariya, 2009; Hanushek, 2013; Amir, Khan, & Bilal, 2015; Dissou, Didic, & Yakautsava, 2016). Instead, the growth might be treated as a function of both the demand, i.e.- job characteristics in which how workers are assigned in their jobs and supply side (attained education).

References

1. Abdullah, A. J., Doucouliagos, H., & Manning, E. (2014). Are regional incomes in Malaysia converging? Papers in Regional Science, 94, S69–S94. https://doi.org/10.1111/pirs.12105

2. Acemoglu, D., & Angrist, J. (2001). How large are human-capital externalities? Evidence from compulsory-schooling laws. In NBER Macroeconomics Annual 2000, Volume 15 (pp. 9–74). MIT Press.

3. Adrian, N., Desislava, N., Ganev, P., & Aleksiev, Y. (2018). European Economic and Social Committee

9_{Sicherman and Galor (1991) have observed two stylized facts of over and undereducation with respect to}

earnings outcomes: -

i. Workers in occupations that require less schooling than they actually have (overeducated) earn lower wages than workers with similar levels of schooling who hold jobs that require the level of schooling they have obtained. These overeducated workers, however, earn more than their co- workers who are not overeducated (i.e., who have the required and, therefore, lower schooling).

ii. Workers in jobs that require more schooling than they have obtained (undereducated) receive higher wages than workers with the same level of schooling who work in jobs that require just their level of schooling. Undereducated workers, however, receive lower earnings that do their co-workers with the required and, therefore, higher schooling.

(10)

Skills Mismatches-An Impediment to the Competitiveness of EU Businesses. https://doi.org/10.2864/448258

4. Amir, H., Khan, M., & Bilal, K. (2015). Impact of educated labor force on Economic growth of Pakistan: A human capital perspective. European Online Journal of Natural and Social Sciences, 4(4), 814–831.

5. Athukorala, P. chandra, & Narayanan, S. (2018). Economic corridors and regional development: The Malaysian experience. World Development, 106, 1–14. https://doi.org/10.1016/j.worlddev.2018.01.009 6. Battu, H., Belfield, C. R., & Sloane, P. J. (2004). Human capital spillovers in the workplace: evidence

for the service sector in Britain. International Journal of Manpower, 25(1), 123–138. https://doi.org/10.1108/01437720410525036

7. Battu, Harminder, Belfield, C., & Sloane, P. (2003). Human Capital Spillovers within the Workplace: Evidence for Great Britain*. Oxford Bulletin of Economics, 65(5), 0575–0594. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.1468-0084.2003.00062.x/full

8. Belfield, C. (2010). Over-education: What influence does the workplace have? Economics of Education Review, 29(2), 236–245. https://doi.org/10.1016/j.econedurev.2009.08.001

9. Benhabib, J., & Spiegel, M. (1994). The role of human capital in economic development evidence from aggregate cross-country data. Journal of Monetary Economics, 34, 143–173.

10. Breton, T. R. (2011). The quality vs. the quantity of schooling: What drives economic growth? Economics of Education Review, 30(4), 765–773. https://doi.org/10.1016/j.econedurev.2011.01.005 11. Büchel, F. (2002). The effects of overeducation on productivity in Germany - The firms’ viewpoint.

Economics of Education Review, 21(3), 263–275. https://doi.org/10.1016/S0272-7757(01)00020-6 12. Büchel, F., & Battu, H. (2003). The Theory of Differential Overqualification : The Theory of

Differential Overqualification : Does it Work ? Scottish Journal of Political Economy, 50(511), 1–16. 13. Cabus, S. J., & Somers, M. A. (2018). Mismatch between education and the labour market in the

Netherlands: is it a reality or a myth? The employers’ perspective. Studies in Higher Education, 43(11), 1854–1867. https://doi.org/10.1080/03075079.2017.1284195

14. Chiswick, B. R., & Miller, P. W. (2010). Does the choice of reference levels of education matter in the ORU earnings equation? Economics of Education Review, 29(6), 1076–1085. https://doi.org/10.1016/j.econedurev.2010.06.001

15. Di Paolo, A., & Mañé, F. (2016). Misusing our talent? Overeducation, overskilling and skill underutilisation among Spanish PhD graduates. Economic and Labour Relations Review, 27(4), 432– 452. https://doi.org/10.1177/1035304616657479

16. Dissou, Y., Didic, S., & Yakautsava, T. (2016). Government spending on education, human capital

accumulation, and growth. Economic Modelling, 58, 9–21.

https://doi.org/10.1016/j.econmod.2016.04.015

17. Fleming, C. M., & Kler, P. (2008). I’m too clever for this job: a bivariate probit analysis on overeducation and job satisfaction in Australia. Applied Economics, 40(9), 1123–1138. https://doi.org/10.1080/00036840600771254

18. Fu, S. (2007). Smart Café Cities: Testing human capital externalities in the Boston metropolitan area. Journal of Urban Economics, 61(1), 86–111. https://doi.org/10.1016/j.jue.2006.06.002

19. Groot, W., & van den Brink, H. M. (2007). Overeducation in the labour market. In Human Capital:

Advances in Theory and Evidence (Vol. 20, pp. 101–112).

https://doi.org/10.1017/CBO9780511493416.008

20. Guironnet, J.-P., & Peypoch, N. (2007a). Human capital allocation and overeducation: A measure of

French productivity (1987, 1999). Economic Modelling, 24(3), 398–410.

21. Guironnet, J.-P., & Peypoch, N. (2007b). Human capital allocation and overeducation: A measure of

French productivity (1987, 1999). Economic Modelling, 24(3), 398–410.

22. Gyimah-Brempong, K., Paddison, O., & Mitiku, W. (2006). Higher education and economic growth in Africa. Journal of Development Studies, 42(3), 509–529. https://doi.org/10.1080/00220380600576490 23. Hanushek, E. A, & Wößmann, L. (2010). Education and Economic Growth. In P. Peterson, E. Baker, &

(11)

https://doi.org/10.2307/2297229

24. Hanushek, Eric A. (2013). Economic growth in developing countries: The role of human capital. Economics of Education Review, 37, 204–212. https://doi.org/10.1016/j.econedurev.2013.04.005 25. Hanushek, Eric A., Jamison, D. T., Jamison, E. A., & Woessman, L. (2008). Education and Economic

Growth: It’s not just going to school, but learning there while there that matters. Education Next, 8(2), 62–71.

26. Hanushek, Eric A, Schwerdt, G., Wiederhold, S., & Woessmann, L. (2015). Returns to skills around the

world: Evidence from PIAAC. European Economic Review, 73, 103–130.

https://doi.org/10.1016/j.euroecorev.2014.10.006

27. Hartog, J. (2000). Over-education and earnings: Where are we, where should we go? Economics of Education Review, 19(2), 131–147. https://doi.org/10.1016/S0272-7757(99)00050-3

28. Hartog, J., & Oosterbeek, H. (1988). Education, allocation and earnings in the Netherlands: 0verschooling? Economics of Education Review, 7(2), 185–194. https://doi.org/10.1016/0272-7757(88)90043-X

29. Hensen, M. M., de Vries, M. R., & Cörvers, F. (2009). The role of geographic mobility in reducing education-job mismatches in the Netherlands. Papers in Regional Science, 88(3), 667–682. https://doi.org/10.1111/j.1435-5957.2008.00189.x

30. Hutchinson, F. E. (2017). Evolving Paradigms in Malaysia’s Regional Development Policy. Journal of Southeast Asian Economies, 34(3), 462–487. https://doi.org/10.1355/ae34-3c

31. Jaoul-Grammare, M., & Guironnet, J. P. (2009). Does over-education influence French economic growth? Economics Bulletin, 29(2), 1190–1200.

32. Jones, M. K., Jones, R. J., Latreille, P. L., & Sloane, P. J. (2009). Training, Job Satisfaction, and Workplace Performance in Britain: Evidence from WERS 2004. Labour, 23(2009), 139–175. https://doi.org/10.1111/j.1467-9914.2008.00434.x

33. Kampelmann, S., & Rycx, F. (2012). The impact of educational mismatch on firm productivity: Evidence from linked panel data. Economics of Education Review, 31(6), 918–931.

34. Kulkarni, M., Lengnick-Hall, M. and Martinez, P. (2015), "Overqualification, mismatched qualification, and hiring decisions: Perceptions of employers", Personnel Review, Vol. 44 No. 4, pp. 529-549. https://doi.org/10.1108/PR-11-2013-0204

35. Leuven, E., & Oosterbeek, H. (2011). Overeducation and mismatch in the labor market. Handbook of the Economics of Education Review, (5523), 1–56. https://doi.org/http://dx.doi.org/10.1016/B978-0-444-53444-6.00003-1

36. Liu, Z. (2007). The external returns to education: Evidence from Chinese cities. Journal of Urban Economics, 61(3), 542–564. https://doi.org/10.1016/j.jue.2006.08.007

37. Lucas, R. E. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(February), 3–42. https://doi.org/10.1016/0304-3932(88)90168-7

38. Mahy, B., Rycx, F., & Vermeylen, G. (2015). Educational Mismatch and Firm Productivity : Do Skills , Technology and Uncertainty Matter ? De Economist, 163(2), 233–262. https://doi.org/10.1007/s10645-015-9251-2

39. Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A Contribution to the Empirics of Economic Growth. The Quarterly Journal of Economics, 107(2), 407–437. https://doi.org/10.2307/2118477

40. Mavromaras, K., Mcguinness, S., O’Leary, N., Sloane, P., & Fok, Y. K. (2010). the Problem of Overskilling in Australia and Britain. The Manchester School, 78(3), 219–241. https://doi.org/10.1111/j.1467-9957.2009.02136.x

41. Mcgoldrick, K., & Robst, J. (1996). Gender differences in overeducation: a test of the theory of differential overqualification. The American Economic Review, 86(2), 280-284.

42. McGowan, M. A., & Andrews, D. (2017). “Labor Market Mismatch and Labor Productivity: Evidence from PIAAC Data,” Research in Labour Economics, in: Skill Mismatch in Labor Markets., 45, 199-241 Emerald Publishing Ltd.

43. McGuinness, S., & Sloane, P. J. (2011). Labour market mismatch among UK graduates: An analysis

using REFLEX data. Economics of Education Review, 30(1), 130–145.

https://doi.org/10.1016/j.econedurev.2010.07.006

(12)

45. Mohamed Noor, N. F., Zakariya, Z., Rambeli, N., & Razak, A. A. (2017). Externalities of Colleagues’ Human Capital Accumulation and Individual Wages: Empirical Evidence from the Malaysian Service Sector. International Journal of Human Resource Studies, 7(1), 192–212. https://doi.org/10.5296/ijhrs.v7i1.10789

46. Moretti, E. (2004a). Human capital externalities in cities. Handbook of Regional and Urban Economics,

4(April), 2243–2291. Retrieved from

http://www.sciencedirect.com/science/article/pii/S1574008004800087

47. Moretti, E. (2004b). Workers’ education, spillovers, and productivity: evidence from plant-level production functions. American Economic Review, 94(3), 656–690. Retrieved from http://www.jstor.org/stable/3592947

48. Nelson, R. R., & Phelps, E. S. (1966). Investment in humans, technological diffusion, and economic growth. The American Economic Review, 69–75.

49. Philipp, G. (2016). The impact of overeducated and undereducated workers on establishment-level productivity: First evidence for Germany. International Journal of Manpower, 37(2), 372–392.

50. Ragayah, H. M. Z. (2008). Income Inequality in Malaysia. Asian Economic Policy Review, 3(1), 114– 132. https://doi.org/10.1111/j.1748-3131.2008.00096.x

51. Ramos, R., Surinach, J., & Artís, M. (2012). Regional Economic Growth and Human Capital: The Role

of Over-education. Regional Studies, 46(10), 1389–1400.

https://doi.org/10.1080/00343404.2012.675140

52. Romer, P. M. (1990). Endogenous Technological Change. Journal of Political Economy, 98(5), 71–102. https://doi.org/10.1093/acprof:osobl/9780199663897.003.0004

53. Romer, P. M. (1994). The Origins of Endogenous Growth. Journal of Economic Perspectives, 8(1), 3– 22. https://doi.org/10.1257/jep.8.1.3

54. Sánchez-Sánchez, N., & Mcguinness, S. (2013). Decomposing the impacts of overeducation and overskilling on earnings and job satisfaction: An analysis using REFLEX data. Education Economics, 23(4), 419–432. https://doi.org/10.1080/09645292.2013.846297

55. Sand, B. M. (2013). A re-examination of the social returns to education: Evidence from U.S. cities. Labour Economics, 24, 97–106. https://doi.org/10.1016/j.labeco.2013.07.001

56. Sicherman, N., & Galor, O. (1991). Overeducation in the labor market. Journal of Labor Economics, 9(2), 101–122.

57. Sloane, P. J., Battu, H., & Seaman, P. T. (1996). Overeducation and the formal education/experience

and training trade-off. Applied Economics Letters, 3(8), 511–515.

https://doi.org/10.1080/135048596356131

58. Tsang, M. C. (1987). The impact of underutilization of education on productivity: A case study of the US Bell companies. Economics of Education Review, 6(3), 239–254.

59. Verhaest, D., & Omey, E. (2006). The Impact of Overeducation and its Measurement. Social Indicators Research, 77(3), 419–448. https://doi.org/10.1007/s11205-005-4276-6

60. Verhaest, D., & Verhofstadt, E. (2016). Overeducation and job satisfaction : the role of job demands and control. International Journal of Manpower, 37(3), 456–473. https://doi.org/10.1108/IJM-04-2014-0106

61. Yussof, I., & Kasim, M. Y. (2003). HUMAN RESOURCE DEVELOPMENT AND REGIONAL COOPERATION WITHIN BIMP-EAGA: ISSUES AND future directions. Asia-Pacific Development Journal, 10(2), 41–56.

62. Yussof, I., & Zakariya, Z. (2009). Economic growth and higher education demand in Malaysia. Jurnal Ekonomi Malaysia, 43(1).

63. Zakariya, Z. (2013). Returns to Education : What Does Over-education Play. Journal of Contemporary Issues and Thought, 3, 138–155.

64. Zakariya, Z., Abdul Jalil, N., & Yin Yin, K. (2017). The Incidence and the Effect of Overskilling on Individuals ’ Wages in Malaysia : A Quantile Regression Approach. Jurnal Ekonomi Malaysia, 51(1), 41–56. https://doi.org/10.17576/JEM-2017-5001-4

65. Zakariya, Z., Abdul, N., Khoo, J., Yin, Y., Fazlin, N., & Noor, M. (2017). The Incidence and the Effect of Overskilling on Individuals’ Wages in Malaysia: A Quantile Regression Approach (Insiden dan Kesan Terlebih Kemahiran ke atas Upah Individu di Malaysia: Satu Pendekatan Regresi Kuantil).

(13)

Jurnal Ekonomi Malaysia, 51(1), 39–54. https://doi.org/10.17576/JEM-2017-5001-4

66. Zakariya, Z., & Battu, H. (2013). THE EFFECTS OF OVEREDUCATION ON MULTIPLE JOB SATISFACTION TOWARDS ENHANCING INDIVIDUAL S ‘ WELL-BEING IN MALAYSIA. Business and Management Quarterly Review, 4, 38–51.

67. Zakariya, Z., Hermanssons, K., Yin, K. Y., Fazlin, N., & Noor, M. (2019). Regional Economic Growth in Malaysia : Does Aggregate Overqualification Matter ?, 10(5), 139–156. https://doi.org/10.5430/rwe.v10n5p139

68. Zakariya, Z., & Yin, K. Y. (2017). Over- and Underskilling in the Malaysian Labour Market: Evidence from the 2003 - 2012 Labour Force Survey (LFS). International Journal of Academic Research in Business and Social Sciences, 6(11), 203–223. https://doi.org/10.6007/ijarbss/v6-i11/2386