Impact of employment protection legislation on employment and exporting in select African countries

Abstract

Labour market flexibility is an important issue in both development and labour economics. More flexibility in the labour market is believed to facilitate job creation, but also makes it easy for employers to terminate employment contracts and may be in conflict with the notion of decent jobs as promoted by the International Labour Organization and workers' unions. It is therefore not surprising that labour market flexibility or inflexibility has received a lot of attention in the extant literature. Using a sample of about 4700 firms from six African countries, we investigate the impact of restrictive labour regulation on a number of economic outcomes and find that more restrictive labour market regulations are detrimental to export propensity, export intensity, investment and employment. Policy-makers must be cautious, however, when implementing employment regulations as too flexible regulations may benefit employers at the expense of employees.

Keywords:

• employment protection
• labour market flexibility
• African firms
• regulation
• exports
• employment
• investment

1. Introduction

Until recently, many African economies were characterised by stunted economic growth as well as high levels of poverty and unemployment (Collier & Gunning, 1999; Elbadawi et al., 2006). A lot of research effort has thus been expended trying to better understand the factors contributing to the continent's poor performance (see for example Rodrik, 1998; Elbadawi et al., 2006; Sachs & Warner, 1997). Among the factors found to be behind Africa's poor economic performance are low levels of foreign direct investment, poor export promotion policies, infrastructure inadequacies, civil conflict and poor governance (Rodrik, 1998; Elbadawi et al., 2006; Sachs and Warner, 1997). However, other researchers argue that labour market flexibility³ is also one of the prerequisites for economic growth, especially through its effect on foreign direct investment inflows.⁴ Another strand of literature also argues that highly restrictive employment laws tend to reduce a firm's likelihood to participate in export markets (Bernard et al., 2007; Seker, 2012). Even though the adverse effects of labour market regulations are acknowledged in the extant literature, there are also a number of studies which argue that employment protection laws (EPLs) are also welfare enhancing and can even improve firm productivity (Benjamin, 2005). Furthermore, in the absence of worker protection, workers may be exploited and employers may decide not to train the workers, thereby adversely affecting firm productivity and the country's competitiveness.

There is therefore no consensus on whether EPLs are beneficial or harmful to the economy. This study seeks to fill this gap by investigating the impact of EPLs on the firm's decision to invest, on export as well as on employment. Most studies trying to investigate the impact of EPLs have tended to focus on developed countries but in this study we depart by looking at a sample of firms from six African countries.

African countries are interesting to study for a number of reasons. Firstly, most African countries have very high levels of unemployment. If indeed EPLs contribute to more unemployment, it is important to compare the benefits and costs of EPLs and see whether the benefits outweigh the costs. If the costs are too high, African governments may need to reconsider and adopt policies that reduce unemployment. Secondly, most African countries do not have comprehensive social protection systems. In the absence of safety nets that cushion the society's most vulnerable from the ravages of poverty, EPLs may be the best mechanism to protect the workers who would have been forced into long spells of unemployment. Lastly, in the absence of EPLs employers may take advantage of workers with weak bargaining power and offer them unfair contracts.

This paper is organised as follows. Section 2 covers both theoretical and empirical review of the literature, whilst Section 3 covers the methodology and descriptive analysis of data. The analysis of results, robustness checks and conclusions are respectively covered in Sections 4, 5 and 6.

2. Literature review

EPLs, which are broadly defined as those regulations that govern the hiring and firing of workers, reduce labour market flexibility and increase labour market inflexibility. There are many adverse effects of EPLs such as reducing turnover in the labour market, limiting job opportunities for new entrants (mostly youths), and driving firms from hiring workers formally to hiring informally. By reducing the turnover in the labour market, EPLs may essentially become barriers of entry for the labour market's new entrants, most of whom are youths with no work experience. Low turnover in the labour market can also increase the duration of unemployment for the youth or those who may have lost their jobs. The impact of low turnover in the labour market may be significant for the youth, especially given that high durations of unemployment may permanently affect their long-term career prospects and trajectories. Even though Howell et al. (2007) recognise that an expanding body of empirical literature contends that employment-unfriendly labour market institutions explain the pattern of unemployment across many countries, their results refute this widely accepted claim. They found little evidence to suggest that 1990s reforms of core protective labour market institutions can explain much of either the success of the ‘success stories’ or the continued high unemployment in many European countries.

According to Botero et al. (2004), labour market rigidity⁵ is driven by three main factors: rigidity in worker hiring, rigidity in workers' working hours, and rigidity in firing workers. Hiring may be made difficult by EPLs that prohibit employers from employing workers on fixed-term contracts (especially by favouring permanent contracts). This may make it difficult for the employers that want to hire workers on short-term contracts or those that may have uncertainty as to the demand for their products. Labour market rigidity can also come in the form of EPLs that have restrictions on night work, holiday work and restrictions on the number of work days per week as well as the number of working hours per day (Botero et al., 2004). EPLs can also increase labour market rigidity by making it difficult for employers to make workers redundant. This may include requiring employers to notify and obtain government approval before firing a worker. Employers may also be required to retrain or reassign workers before making them redundant (Botero et al., 2004). Opponents claim that flexibility puts all of the power in the hands of the employer, resulting in an insecure workforce, and that this may affect their productivity. Botero et al. (2004) categorised the above EPLs into three indices, which they then used to construct an index of employment rigidity. The employment rigidity index is an average of the three indices. It ranges from 0 to 100, with 0 indicating the absence of any EPLs or any rigidity and 100 indicating a very rigid labour market (Botero et al., 2004).

According to the World Bank's Doing Business database, Africa has the highest number of employment regulations compared with other regions (African Economic Outlook, 2012; World Bank, 2012). The 2008 average employment rigidity indicator for the sub-Saharan Africa region was as high as 47 whilst that of the other regions of the world was on average below 32. One must however note that enacting good laws is one thing but implementing them properly is something else. African governments are well known for not effectively implementing the regulations that they enact. The poor enforcement of regulations may explain why labour regulations are not considered to be a significant obstacle to the operations of most firms in Africa. Most firms in the continent consider poor access to finance and access to infrastructure (such as roads, electricity and water) as important obstacles rather than labour regulations. The distortionary impact of labour regulations in Africa, however, is an important issue to investigate, especially given the continent's high level of unemployment. For example, the crisis in Tunisia that resulted in the government being changed was instigated by unemployed youths.⁶ In South Africa a significant number of the unemployed is made up of the youth with just high school qualification. Given the high number of employment regulations in South Africa, it is possible that such unemployment may be partly due to EPLs that make it difficult for employers to hire and fire workers. This also follows the notion that ‘if it is difficult to fire do not hire’. This view is partly supported by Burges & Fintel (2009), who argue that higher unemployment rates faced by the young are predominantly due to the disadvantage of entering the labour market, rather than being attributable to their age. However, Banerjee et al. (2008) ascribe unemployment levels to changes in the structure of production as well as skills-biased technological change. They argue that unemployment is a result of the demand for unskilled or semi-skilled workers (who are in abundant supply) decreasing, whilst demand for highly skilled workers (for whom the unemployment rate is already low) is increasing. Even though other researchers have argued that the cause for the slow growth in employment could be ascribed to the enactment of tighter labour market legislation that endowed trade unions with greater bargaining power and increased the coverage of minimum wages, Woolard & Woolard (2005) and Banerjee et al. (2008) all conclude that the increase in unemployment in South Africa coincides with a decrease in the real wages of unskilled and semi-skilled workers, making it unlikely that wage increases could have contributed to the rising unemployment rate. However, Banerjee et al. (2008: 717) still believe that the labour market is characterised by ‘institutional constraints that keep wages from declining as much as they could [in the face of a large surplus of unskilled workers]’. The truth, however, is that there are so many factors that have been put forward to explain the causes of high unemployment in South Africa (historical, social and economic factors), but the main aim of this study is to empirically ascertain the impact of labour market regulations on firm size, investment and the likelihood by firms to enter export markets in Africa. This is because firm size (employment), investment and exporting are aspects that influence growth and hence employment creation. This study is also a contribution to the growing debate about Africa's growth challenges.

While the evidence on the impact of labour regulations seems to overwhelmingly suggest that they are bad for the workers, firms and the economy in general, Almeida (2008) suggests that there are both positive and negative effects of labour regulations. He argues that the benefits and costs of regulations must be compared to see which of the two outweigh the other. For example, on the link between EPLs and firm size or firm productivity, Almeida (2008) argues that labour market regulations tend to increase the cost of hiring workers, forcing firms to hire fewer workers. This may affect firm size, especially in cases where employment is measured using the number of workers. Using a sample of firms from 63 countries Almeida (2008) found that firms operating in environments or countries where labour markets are highly regulated tend to be smaller in size. This is the case especially in labour-intensive sectors of the economy. The evidence on this relationship is weak in countries with poor governance, however, suggesting that countries with poor legal environments may be failing to efficiently implement labour regulations.

Almeida (2008) further argues that EPLs can also affect firm productivity, affecting firm size (when firm size is measured using sales or turnover). If firm size and productivity are affected this may have huge implications on the entire economy, as the economy may fail to absorb the new labour market entrants such as the youths (Almeida, 2008). A country's foreign currency earnings as well as its balance of payments position may be affected if the decrease in production affects the firm's exports. Almeida also argues that strict EPLs may reduce turnover in the labour market; with firms being forced to restrict firing during bad economic conditions, like recessions, and not optimally hiring during good economic conditions. Less hiring during good economic conditions is usually done by firms to avoid high future firing costs. The evidence on the adverse impact of EPLs on firm size is further corroborated by Almeida & Carneiro (2009), who used data on Brazilian firms. Agell (1999) also argues that firms believe that the price tag on wage flexibility is quite high. Highly flexible labour markets are more likely to witness wage decreases. But employers may be unwilling to cut wages because they fear wage cuts will affect morale in an adverse manner. Some managers are concerned that employees may interpret a wage cut as an unfriendly act, which will affect work effort and the quality of future job applicants. Agell (1999) argues that lower wages will lead to less effort and will cause workers to lose their allegiance to the organisation. Agell (1999) goes on to argue that reciprocity appears important for the workings of the labour market. A firm that raises the wage induces a positive effort response, whilst the firm that cuts wages provides workers with an incentive to retaliate. Keynes (1936) in his general theory also supported this by arguing that inflexible wages and unemployment are phenomena of an essentially social nature and that wage competition is socially improper. This norm works against wage flexibility and the market forces of supply and demand as well as labour market flexibility (Agell, 1999).

Further evidence on the relationship between labour regulation and economic outcomes is found in Besley & Burges (2004). Using data on Indian firms, Besley & Burges (2004) found that high labour market rigidities adversely affect employment, investment and firm productivity. Hopenhayn & Rogerson (1993) also demonstrate that highly restrictive labour laws can affect firm creation as well as employment creation. This is further supported by Kaplan (2009), who found that flexible labour markets, through job reallocation, tend to increase employment creation. The gains of labour market flexibility are more pronounced in those environments or countries with highly regulated labour markets. Siebert's (1997) prominent paper in the Journal of Economic Perspectives aptly summed up the conventional wisdom: ‘Labor Market Rigidities: At the Root of Unemployment in Europe‘. Therefore countries with high levels of unemployment were repeatedly urged to undertake comprehensive structural reforms to reduce labour market rigidities (Howell et al., 2007).

Botero et al. (2004) argue that countries with highly regulated labour markets tend to have bad labour market outcomes (such as high levels of unemployment and low labour market participation rates). For example, Amin (2009) found that labour market rigidity adversely affects employment in India. Heckman & Pages (2000) found that labour market regulation not only affects employment but also tends to worsen the problems of inequality. Ahsan & Pages (2009) as well as Kugler (2004) argue that labour market regulation tends to reduce employment and output; a fact further supported by Feldman (2008).

Highly restrictive employment laws can also adversely affect trade; especially a firm's decision to export. Labour market flexibility, by reducing adjustment costs and increasing productivity, is a source of comparative advantage (Helpmann & Itskhoki, 2010). Higher profits associated with labour market flexibility can facilitate a firm's entry into foreign markets (Helpmann & Itskhoki, 2010). This is especially true if future exporters must first grow in size and be more productive before entering and sustainably participating in foreign product markets (Helpmann & Itskhoki, 2010; Seker, 2012). All this is made impossible by highly restrictive labour laws that tend to reduce firm size, productivity and firm competitiveness (Helpmann & Itskhoki, 2010).

Howell et al. (2007) document a number of studies that find little or no evidence of the orthodox view that rigid labour markets are bad for employment creation. Using cross-country regressions, Nickell's (1997) influential paper finds the EPL variable insignificant. He concludes by arguing that ‘It is clear that the broad-brush analysis that says that European unemployment is high because European labour markets are “rigid” is too vague and probably misleading’ (Nickell, 1997: 73). Using the same data as and newer versions of institutional variables compared with those used by Nickell (1997), Baker et al. (2005) find similar results on the EPL variable. Of the 10 papers reviewed by Howell that were written between the years 1996 and 2006, only three find EPL to be statistically significant at the 5% level. It also appears that the results are very sensitive to model and variable specification. Thus the use of alternative, arguably much superior, OECD-generated measures of unemployment benefit replacement rates and EPLs actually weakens the results (Howell et al., 2007). Moreover, studies for the United Kingdom by Henry & Nixon (2000) conclude that the time-series evidence suggests that the deregulation of the UK labour market does not account for much of the observed changes in unemployment since the 1970s.

Thus even though theory and some empirical studies support the view that labour market rigidity is not good for employment creation, there is also a large number of studies which have found that the regression evidence in some countries, mostly in Europe, does not offer compelling statistical evidence that protective labour market institutions are at the root of persistent high unemployment. These mixed results basically imply that the impact of EPL on economic variables is not conclusive and robust. This study attempts to find the nature of the relationship between these variables in Africa.

3. Data and methodology

The data used in the paper came from the World Bank's Investment Climate Surveys, which use a common survey instrument across all countries covered. The use of a common survey instrument across all firms in different countries is an important advantage of this dataset. This is because the differences across countries are not due to the differences in data collection techniques or the sampling frame (Kaplan, 2009).⁷

The survey instrument also contains an important question that is important for the construction of the worker-flow or job allocation variables. The firms were asked the following question: ‘In the previous year, did labour regulations affect your decisions about hiring or firing permanent employees in a significant way?’ Firms that said yes to the question were then further asked how many workers they would hire or fire if they were not to comply with the labour regulations. In line with the above question we categorise firms into the following groups: those stating that employment regulations affected only their firing decisions, those stating that employment regulations affected hiring decisions only, those stating that employment regulations affected both hiring and firing, and those stating that neither hiring nor firing decisions were affected.

Job reallocation is given by the summation of the number of people that would have been hired and those that would have been fired if firms did not have to comply with employment regulations. It is agreed in the literature that stricter employment protection leads to lower rates of job creation (people hired) and job destruction (people fired) (see Addison & Teixeira, 2001). Whether stricter employment regulations result in lower employment or higher unemployment remains an empirical question (Addison & Teixeira, 2001). Since job reallocation gives us the extent to which firms have the freedom to hire or fire in the absence of labour regulations, we use it to proxy for employment regulation. For example, relaxing the labour regulation makes it easy for firms to respond to shocks (e.g. product demand shocks, increases in wages, or changes in the price of labour substitutes, etc.) by hiring or/and firing more workers. Relaxing labour regulation therefore increases job reallocation (or resource allocation), while a more restrictive ER reduces job reallocation. For purposes of this study we therefore argue that an increase in job reallocation implies a relaxation of the employment regulations.

3.1 Methodology

To better understand the impact of employment regulation on different economic outcomes such as investment, employment, decision to export and export intensity, we use the following specification:

where Y_i stands for the economic outcome, ER_i stands for employment protection regulation indicator, and X_i stands for the control variables that also affect the indicated economic outcomes. The control variables include foreign ownership, age, innovation, sales (profitability), and location. Table A1 in Appendix A presents a more detailed description of all variables used in the study.

More specifically, the estimated regressions are as follows:

Employment regulation can also affect the decision to export goods and or services. To capture this relationship we ran a probit model. Firms were asked whether they directly or indirectly exported some of their products/services. An export dummy taking a value of 1 if the firm export and 0 otherwise was created to capture export propensity. The model is estimated by regressing the probability of being an exporter as a function of firm characteristics as follows:

3.2 Data description

Table 1 provides the descriptive statistics for the data used for the study. The average age for all firms in all countries is 15.15 years. About one-quarter of the firms were located in the capital city. When it comes to ownership, Botswana had the highest percentage of foreign-owned firms, and Nigeria had the lowest proportion. About 47% of the firms in Botswana were foreign owned, while in Nigeria only 2% were foreign owned. Botswana had the largest number of firms (67%) that stated they introduced a new product or a significantly improved product into the market. Kenya had the lowest, with 35% of the firms stating that they introduced a new product into the market. Swaziland, at 37%, had the largest number of firms that directly exported its products or services, whilst Nigeria had the lowest, at 1%. The number of firms in the sample countries ranged from 70 in Swaziland to 3157 in Nigeria. The total number of firms from all sampled countries is 4702. For the whole sample of firms, the average firm's size is 39.01, with smallest firms having a size of one employee and the largest having a size equalling 4000 workers. Swaziland, at 26.47% of sales, has the largest export intensity and Nigeria at 0.47% of sales has the smallest mean export intensity. This is not surprising since a significant number of Nigerian firms stated that they do not export goods or services. The proportion of Nigerian exporting firms is only 1.3%.

Table 1: Descriptive statistics

Variable		Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Employment / Size	Mean	66.24	79.34	23.23	183.77	51.99	51.64	39.01
	SD	128.59	239.90	97.57	395.19	108.39	239.53	154.38
	Minimum	5	1	1	5	5	5	1
	Maximum	1 000	3 000	3 000	3 000	1 118	4 000	4 000
Export Intensity	Mean	8.486	6.19	0.47	26.47	2.34	4.42	2.369
	SD	23.325	17.12	4.72	39.88	8.86	16.82	11.731
	Minimum	0	0	0	0	0	0	1
	Maximum	100	100	100	100	75	100	100
Job Reallocation	Mean	3.404	1.045	0.338	16.129	0.418	0.010	0.748
	SD	23.971	8.161	7.080	119.45	2.645	0.171	16.499
	Minimum	0	0	0	1	0	0	0
	Maximum	250	150	300	1 000	34	3	1 000
Foreign	Mean	0.474	0.110	0.015	0.386	0.139	0.176	34.074
	SD	0.502	0.313	0.121	0.490	0.347	0.381	128.157
	Minimum	0	0	0	0	0	0	0
	Maximum	1	1	1	1	1	1	4 000
Innovate^a	Mean	0.667	0.352		0.586	0.648	0.596	0.160
	SD	0.473	0.478		0.496	0.478	0.491	0.367
	Minimum	0	0		0	0	0	0
	Maximum	1	1		1	1	1	1
Age	Mean	11.342	16.105	15.278	11.779	14.277	12.828	15.152
	SD	8.515	14.571	10.348	10.803	12.088	10.124	11.258
	Minimum	1	1	2	1	1	1	1
	Maximum	37	87	158	47	98	79	158
Sales	Mean	7 765 031	4.02 × 10^8	8.71 × 10^7	1.51 × 10^7	1.56 × 10^9	1.99 × 10^9	1 544 166
	SD	1.95 × 10^7	3.39 × 10^9	7.40 × 10^7	3.20 × 10^7	8.94 × 10^9	1.07 × 10^10	2.00e × 10^7
	Minimum	0	40 000	0	980 000	650 000	1 000 000	0
	Maximum	1.72 × 10^8	7.50 × 10^10	1.90 × 10^10	1.90 × 10^8	1.35 × 10^11	1.36 × 10^11	1.11 × 10^9
Location	Mean	0.807	0.597	0.050	0.214	0.689	0.818	0.249
	SD	0.396	0.491	0.218	0.413	0.464	0.387	0.432
	Minimum	0	0	0	0	0	0	0
	Maximum	1	1	1	1	1	1	1
Export Propensity	Mean	0.193	0.201	0.013	0.371	0.128	0.124	0.068
	SD	0.396	0.401	0.115	0.487	0.335	0.330	0.252
	Minimum	0	0	0	0	0	0	0
	Maximum	1	1	1	1	1	1	1

Source: Authors' own calculations from the Investment Climate Assessment survey data.

Notes: SD, standard deviation. Export is a dummy variable equal to 1 if the firm directly exports some of its products/services. Export Intensity is the percentage of sales gained from exports. Foreign is a dummy variable equal to 1 if more than 10% of the firm is owned by foreign individuals or entities. Innovate is a dummy variable equal to 1 if the firm introduced a new product into the market or significantly improved a product/service in the last three years. Age is a firm's age and is calculated by subtracting the year the firm started operations from the survey year. Location is a dummy variable equal to 1 if the firm is located in the capital city. Obstacle measures the extent to which labour regulations are an obstacle to a firm's operations; it takes values from 1 to 5 (with 1 standing for no obstacle and 5 standing for very severe obstacle). Fire is the number of workers a firm would fire if it did not have to comply with labour regulations. Hire is the number of workers a firm would hire if it did not have to comply with labour regulations. Job reallocation is the summation of Fire and Hire. Investment is the amount spent purchasing machinery, vehicles, equipment, land and building in the previous year (a year prior to the survey year). Employment is the number of employees employed by the firm. ^aThe questionnaire on Nigeria does not have the question on innovation.

As can be seen from Table 2, a significant number of firms in the sampled countries stated that their hiring or firing decisions were not affected by labour regulations. For example, almost all firms in Uganda (99.67%) stated that their decisions were not affected by labour regulations, and 81.43% of the firms in Swaziland stated that their hiring and firing decisions were not affected by the labour regulations. The percentage of firms stating that only their firing decisions were affected by the labour regulations were about 9% of the firms in Swaziland (the highest among the sample countries), 5.26% in Botswana, 2.30% in Kenya and 3.30% in Tanzania. In Uganda the regulations do not seem to affect the firing decisions, while only 0.38% of the Nigerian firms stated that the regulations affected their firing decisions. This seems to suggest that most EPLs in a number of African countries are not being enforced and hence they do not appear to have seriously affected firm behaviour.

Table 2: Percentage of firms whose firing and hiring decisions were affected by employment regulations

Country	Observations	Mean employment	Hiring and firing not affected (%)	Only firing affected (%)	Only hiring affected (%)	Hiring and firing affected (%)	Number fired	Number hired	Net (hire less fire)	Job reallocation
Botswana	114	66.24	82.46	5.26	2.63	11.00	23	365	342	388
Tanzania	272	51.99	95.60	3.30	0.00	0.73	82	32	–50	114
Uganda	307	51.64	99.67	0.00	0.33	0.00	0	3	3	3
Nigeria	2 387	23.23	98.41	0.38	0.80	0.42	545	523	–22	1 068
Swaziland	70	183.77	81.43	8.57	4.29	5.71	68	1 061	993	1 129
Kenya	781	79.34	92.19	2.30	1.92	3.59	343	473	130	816

Source: Authors' own calculations from the Investment Climate Assessment survey data.

On whether employment regulation affected hiring, there is a small percentage of firms that seem to think that only hiring was affected. The highest number of firms who indicated that hiring was affected was recorded in Swaziland, at 4.29%. In Tanzania the hiring decisions of the firms were not affected by the labour regulations, whilst in Botswana and Kenya 2.63% and 1.92% of the firms respectively stated that only hiring was affected. In Uganda and Nigeria the figures were very small, 0.33% and 0.80% respectively.

Another important category from which vital information can be teased is that of firms which felt that both hiring and firing decisions were affected by the employment regulation. About 11% of the firms in Botswana stated that both firing and hiring decisions were affected by the labour regulations, followed by Swaziland with 5.71% and Kenya with 3.59%. Only 0.73% and 0.42% of the firms in Tanzania and Nigeria stated that their firing and hiring decisions were affected by the employment regulation, whilst in Uganda no firms were affected.

When it comes to the actual number of workers that would have been fired or hired by the firms if they were not to comply with the employment regulation, it seems that in many of these sampled countries more people would have been hired than fired. For example, firms in Botswana would have hired 365 more people and fired only 23 workers if they did not have to comply with labour regulations, resulting in a net increase in employment by 342. In Swaziland 1061 more people would have been hired and only 68 would have been fired if firms were not to comply with labour regulations, resulting in 993 more people employed. In Kenya we have 473 more people being hired and 343 being fired, whilst in Uganda three more people would have been hired while nobody would have been fired. Only in Nigeria and Tanzania would the number of people fired be greater than the number of people hired. Table 2 also shows the net change in the number of employees calculated by subtracting the number of people that would have been fired from those that would have been hired. The net worker-flow is highest in Swaziland (993) and smallest in Tanzania (–50). Looking at the absolute figures of the net worker flow we find that Ugandan and Nigerian firms are the least affected by the employment regulation, while Swaziland is the country most affected by employment regulation. One must note that one important impact of the employment regulation is not observed from the data. This is the number of people that would be employed by new firms which could have been established if there were no restrictive labour regulations.

Table 2 also shows the job-reallocation figures for the sample countries. The job reallocation was highest in Swaziland, at 1129 people, and lowest in Uganda, at three people. This result is in line with our expectation that countries such as Swaziland have very restrictive employment regulation, and if firms in these countries do not comply with the labour regulations they would reallocate a significant amount of their resources (in this case labour), resulting in high values of job reallocation figures. This corroborates findings by Kaplan (2009), who found that countries with more restrictive labour regulations or highly regulated markets tend to experience larger gains in total employment when employment regulations are relaxed.

4. Analysis of results

Equations (2) to (5) were estimated for each country. All firms in the sample countries were also pooled and a single cross-country regression model was estimated. The models estimated are all significant as shown by the significant F-statistics and low p-values. For the equations for the employment determinants, the adjusted r-squared ranged from 48% to 52%, quite high for cross-sectional data. For the export decision the pseudo r-squared was in the range 13 to 38%. The r-squared for the investment equation ranged from 10% to 60%. For the export intensity, r-squared ranged from 8% to 46%.The results suggest that employment regulation as proxied through worker flow changes has an impact on economic outcomes such as employment, decision to export, export intensity and investment.

We now look at the impact of employment regulations on each of the economic outcomes. We start by looking at the relationship between employment regulation and employment. Table 3 presents the results. Relaxing labour market regulations increases job reallocation, which tends to increase the number of people employed by the firms. This is found to be the case in Botswana, Kenya, Nigeria and Tanzania. The result was also found to be the same when a pooled regression that included all countries was used. This result suggests that relaxing labour markets makes it easier for firms to hire and fire and thus makes the labour market more fluid. This tends to result in more people being employed, as shown by the positive relationship between job reallocation and employment. This finding corroborates findings by Botero et al. (2004) and Micco & Pages (2007), who find that more restrictive labour regulations tend to increase unemployment rates in India. Other controls used in the regression were also found to be important. For example, innovative firms in Uganda and Kenya were also found to hire more workers. Older firms also tended to have a larger number of employees than young firms, and the relationship between age and employment was positive in all countries, with the exception of Swaziland. We used sales as a crude measure of profitability and found that more profitable firms tend to hire more workers than less profitable ones, as seen from the positive coefficient of the sales variable. This result is maintained in all the individual country regressions, as well as in the pooled regressions. There is a positive relationship between employment and foreign ownership, suggesting that foreign-owned firms tend to have a higher number of workers than locally owned firms.

Table 3: Results for the impact of labour regulation on employment

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.005***	0.005***	0.006***	–0.003	0.035***	–0.325	0.008***
	(3.84)	(2.93)	(2.95)	(–0.07)	(2.97)	(–1.38)	(5.02)
Foreign	0.140	0.108	0.239*	0.060	0.298*	0.221*	0.522***
	(0.70)	(0.85)	(1.86)	(0.20)	(1.97)	(1.85)	(6.79)
Innovate	0.161	0.434***		–0.444	0.047	0.200**
	(0.72)	(4.91)		(–1.46)	(0.51)	(2.30)
Age	0.296*	0.114*	0.079***	–0.302	0.162**	0.279***	0.159***
	(1.96)	(1.91)	(3.899)	(–1.50)	(2.25)	(3.66)	(6.94)
Sales	0.350***	0.396***	0.357***	0.528***	0.383***	0.362***	0.311***
	(6.57)	(17.82)	(36.34)	(4.97)	(14.84)	(15.15)	(38.55)
Location	0.075	0.165**	0.388***	–0.685*	0.046	0.005	–0.003
	(0.36)	(2.03)	(7.67)	(–1.71)	(0.44)	(0.050	(–0.07)
Country dummy^a							0.452***
							(9.94)
Constant	–2.704***	–3.899***	–3.302***	–2.523	–4.527***	–4.622***	–2.780***
	(–3.69)	(–11.83)	(–21.23)	(–1.63)	(–10.990	(–10.66)	(–22.00)
Observations	90	578	2 740	51	246	280	3 985
R-squared	0.48	0.59	0.58	0.58	0.66	0.61	0.52
F()	36.26	157.32	459.70	15.74	97.50	70.71	509.05
P-value	0.000	0.000	0.0000	0.0000	0.0000	0.0000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%. ^a Kenya is the reference country.

The second relationship to be looked at is that between employment regulation and investment. It is difficult, a priori, to state whether restrictive employment regulations have a detrimental effect on investment or not. On the one hand, very restrictive labour regulations – by making it more difficult to fire workers, for example – tend to increase a firm's adjustment costs (Cingano et al., 2010). This may force employers to move away from labour and start using the relatively flexible input (capital), stimulating investment. However, more restrictive employment regulation can strengthen worker's bargaining power, exacerbating the hold-up problem and ultimately resulting in less investment per worker (Cingano et al., 2010). The final impact of employment regulation on investment therefore depends on the increased worker bargaining power that comes with more restrictive employment regulation and the concomitant move towards more flexible inputs that is also associated with restrictive labour regulations. Using firm-level data on firms from 14 European Union countries, Cingano et al. (2010) investigate the impact of employment regulation on investment and productivity and find that more restrictive EPLs tend to reduce both investment and productivity. Employment regulation can be an obstacle to investment decisions because it tends to stifle resource allocation (Cingano et al., 2010).

We find mixed results on the relationship between investment and employment regulation (see Table 4). In Botswana and Nigeria there is a positive and significant relationship between the two, while for Kenya the relationship is negative and also significant. For the other countries the results are not significant. For the pooled regression, the results are also insignificant. As for the other control variables there is a positive relationship between investment and ownership in Nigeria and Uganda. For Swaziland we find a negative relationship. This suggests that in Botswana and Nigeria the more restrictive employment regulation policies tend to increase Investment more (as employers move away from the more inflexible input, labour) than it reduces investment through the hold-up problem. The reverse is true for Kenya. For the other countries the results are insignificant. The results are also insignificant for the pooled regression. There is also a positive relationship between innovation and investment in Kenya. Older Nigerian firms tend to invest less than younger ones. Across all countries, more profitable firms – probably because of the access to cheaper internal funds – tend to invest more. This result is also maintained when all firms are pooled in one regression.

Evidence abounds in the extant literature that exporting firms are more productive, tend to be large, and hire more workers and tend to grow faster than non-exporting firms (Seker, 2012). In Seker (2012) it is shown that more restrictive employment regulations tend to have a detrimental effect on the propensity to export. It is therefore important to also assess whether employment regulations have any impact on the propensity to export. After running export propensity on employment regulation proxy and other controls, we find that EPL is important when it comes to the export decision (see Table 5). In Nigeria and Swaziland a relaxation of the EPL tends to increase the probability to export. The result is maintained when all firms from all sample countries are pooled into one regression. The results, however, suggest that for the other sample countries (Botswana, Kenya, Tanzania and Uganda) employment regulation is not important when it comes to the decision to export goods and services. Other control variables seem to matter when it comes to the export decision. These include foreign ownership (foreign-owned firms are more likely to export goods when compared with locally owned firms), and innovative Kenyan firms are also more likely to export goods and services as compared with those that are not innovative. Older firms are also more likely to export goods. For all firms, more profitable firms are more likely to export goods and services as compared with less profitable ones. Firms located in the capital city are more likely to export goods as compared with those located elsewhere. All the above results are maintained when a pooled regression is run.

Table 4: Results for the impact of labour regulation on investment

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.009***	–0.008*	0.005**	0.026	–5.930	–0.751	0.001
	(4.68)	(–1.77)	(2.47)	(0.38)	(–0.75)	(–0.54)	(0.73)
Foreign	0.444	0.328	0.423*	–1.495***	8.510	1.180*	–0.034
	(0.89)	(1.17)	(1.90)	(–3.00)	(0.82)	(1.68)	(–0.24)
Innovate	–0.449	0.408*		0.661	–4.620	0.398
	(–0.90)	(1.96)		(1.31)	(–0.75)	(0.780
Age	0.081	0.103	–0.130*	–0.261	1.710	0.368	–0.071
	(0.19)	(0.75)	(–1.89)	(–0.960	(1.39)	(0.82)	(–1.25)
Sales	0.432***	0.638***	0.710***	1.031***	3.470**	0.917***	0.778***
	(2.95)	(13.18)	(27.36)	(5.84)	(2.04)	(6.53)	(41.61)
Location	0.177	0.162	0.594***	1.178**	–9.440*	–0.747	0.408***
	(0.26)	(0.84)	(4.72)	(2.24)	(–1.66)	(–1.19)	(4.69)
Country dummy							–0.338***
							(–3.30)
Constant	5.459***	2.981***	2.639***	–3.056	–5.610**	1.690***	1.424***
	(2.79)	(3.89)	(6.49)	(–1.18)	(2.20)	(6.64)	(4.87)
Observations	90	578	2 740	42	247	279	2 125
R-squared	0.21	0.54	0.46	0.60	0.10	0.60	0.57
F()	654.10	59.17	259.11	8.98	1.98	70.71	396.68
P-value	0.0000	0.0000	0.0000	0.000	0.104	0.000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%.

Table 5: Results for the impact of labour regulation on the decision to export

Independent Variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	–0.048	–0.001	0.017***	0.448*	–0.018	0.020	0.005*
	(–0.66)	(–0.14)	(2.66)	(1.65)	(–0.48)	(0.011)	(1.77)
Foreign	0.426	0.463***	0.410	1.449***	0.4910	0.582	0.878***
	(1.12)	(2.66)	(1.50)	(2.68)	(1.74)	(0.530	(7.92)
Innovate	0.252	0.504***		0.132	0.209	0.133
	(0.73)	(3.61)		(0.26)	(0.82)	(0.53)
Age	–0.116	0.251***	0.330**	–0.110	0.330*	0.965**	0.173***
	(–0.39)	(2.76)	(2.08)	(–0.31)	(1.91)	(2.54)	(2.85)
Sales	0.176*	0.265***	0.336***	0.471***	0.218***	0.324***	0.237***
	(1.73)	(6.53)	(7.10)	(3.06)	(3.76)	(5.25)	(11.57)
Location	0.840	0.048	0.460***	–0.855	–0.245	–0.140	0.243***
	(1.58)	(0.36)	(2.47)	(–1.19)	(–1.02)	(–0.53)	(2.95)
Country dummy							0.942***
							(11.53)
Constant	–4.361***	–6.367***	–9.072***	–7.858***	–6.219***	–8.657***	–6.51***
	(–3.46)	(–8.95)	(–10.74)	(–3.05)	(–5.45)	(–6.68)	(–17.62)
Observations	90	578	2 740	67	247	306	3 986
R-squared	0.13	0.24	0.38	0.43	0.19	0.34	0.35
F()	12.7	109.53	106.77	19.52	38.68	46.58	532.75
P-value	0.047	0.000	0.0000	0.003	0.0000	0.0000	0.0000
Log pseudo-likelihood	–35.45	–253.29		–19.81	–81.45	–75.28	–699.53

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%.

When it comes to the relationship between EPL and export intensity, we find that a relaxation of the labour laws tend to increase export intensity in Nigeria and Swaziland (see Table 6). For the other countries the relevant coefficient is insignificant. Also, foreign-owned firms tend to have higher export intensity as compared with locally owned ones. This result is maintained when all firms from the sample countries are pooled in one regression. Kenyan firms that innovate also tend to have higher export intensity. Across all sample countries, more profitable firms tend to have higher export intensity when compared with less profitable firms. This relationship is maintained when all firms from the sample countries are included in one pooled regression. Surprisingly, the results suggest that firms located in the capital city tend to have lower export intensity than those outside the capital city.

Table 6: Results for the impact of labour regulation on export intensity

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	–0.032	–0.072	0.166***	3.351***	–0.315	–2.085	0.074
	(–0.41)	(–1.04)	(14.98)	(3.47)	(–1.51)	(–0.39)	(1.61)
Foreign	–3.764	7.065***	5.665***	21.683**	2.990*	0.595	9.402***
	(–0.97)	(3.83)	(8.54)	(2.09)	(1.77)	(0.22)	(5.73)
Innovate	1.741	3.773***		–1.573	–0.489	–0.784
	(0.43)	(2.72)		(–0.16)	(–0.43)	(–0.42)
Age	–2.554	0.714	0.289**	–16.777**	0.659	0.222	0.008
	(–1.20)	(1.09)	(2.26)	(–2.27)	(1.00)	(0.18)	(0.02)
Sales	4.459***	1.945***	0.482***	7.994***	1.002***	3.210***	1.211***
	(3.86)	(7.07)	(9.15)	(2.75)	(3.68)	(6.11)	(8.27)
Location	–3.734	–3.902***	0.566	–9.185	–1.444	–4.460*	–2.492***
	(–0.74)	(–3.36)	(1.53)	(–0.96)	(–1.26)	(–1.90)	(–3.67)
Country dummy							6.101***
							(6.91)
Constant	–49.550***	–27.548***	–8.224***	–59.151	–17.358***	–52.867***	–18.339***
	(–3.14)	(–6.53)	(–9.69)	(–1.26)	(3.53)	(–5.45)	(–8.43)
Observations	112	780	3 157	51	271	307	3 986
R-squared	0.13	0.18	0.14	0.46	0.08	0.13	0.14
F()	2.61	28.56	102.53	14.08	5.08	8.91	31.52
P-value	0.021	0.0000	0.000	0.000	0.0001	0.0000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%.

5. Robustness checks

Two steps were taken to check for the robustness of our results. First, we included the variable Export as an independent variable in the employment and investment equations. Second, we included interaction terms in the employment and investment equations. The results from the first step are reported in Tables 7 and 8. Comparing the results from Table 7 (which includes the Export variable) with those from Table 3 (which does not include the Export variable) we find that most of the results are similar. The coefficient of Export is positive and significant for almost all countries. This suggests that firms which export their goods and services tend to employ more workers than those which do not export. Table 8 provides results from running investment on all independent variables, including Export. Comparing the results from Table 4 with those from Table 8 we notice that most estimated coefficients are not different. The exports coefficient is also positive and significant for the cases of Uganda and the pooled regressions.

Table 7: Results for the impact of labour regulation on employment (including Export as an independent variable)

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.006***	0.005***	0.036*	–0.044	0.038***	–0.298	0.007***
	(4.01)	(3.04)	(1.95)	(–1.58)	(3.27)	(–1.28)	(3.42)
Foreign	0.093	0.012	0.238	–0.334	0.231	0.171	0.345***
	(0.48)	(0.10)	(1.39)	(–1.28)	(1.37)	(1.42)	(4.81)
Innovate	0.137	0.352***		–0.375	0.035	0.194**
	(0.63)	(4.12)		(–1.37)	(0.38)	(2.24)
Age	0.306**	0.080	0.161***	–0.224	0.138*	0.260***	0.134***
	(2.15)	(1.39)	(4.02)	(–1.25)	(1.94)	(3.43)	(6.01)
Sales	0.331***	0.356***	0.326***	0.363***	0.368***	0.338***	0.286***
	(6.27)	(15.27)	(30.45)	(4.02)	(14.66)	(13.06)	(35.80)
Location	0.001	0.154**	0.032	–0.606	0.060	0.019	–0.013
	(0.01)	(1.98)	(0.72)	(–1.67)	(0.56)	(0.17)	(–0.35)
Export	0.509**	0.632	0.288**	1.390	0.406**	0.341**	0.804***
	(2.08)	(6.24)	(2.09)	(4.75)	(2.34)	(2.37)	(10.08)
Country dummy^a							0.313***
							(6.97)
Constant	–2.452***	–3.236***	–2.940***	–0.651	–4.226***	–4.166***	–2.324***
	(–3.20)	(–9.14)	(–15.81)	(–0.49)	(–10.58)	(–8.84)	(–17.91)
Observations	90	578	1 643	51	246	280	3 985
R-squared	0.50	0.62	0.52	0.70	0.67	0.61	0.55
F()	35.57	148.31	192.71	15.90	85.06	62.43	458.37
P-value	0.000	0.000	0.0000	0.0000	0.0000	0.0000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%.^aKenya is the reference country.

Table 8: Results for the impact of labour regulation on investment (including Export as an independent variable)

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.010***	–0.007	0.025	0.035	–6.340	–0.614	0.001
	(4.50)	(–1.28)	(0.88)	(0.52)	(–0.86)	(–0.45)	(0.59)
Foreign	0.415	0.436	0.261	–1.302	9.400	0.915	–0.092
	(0.84)	(1.61)	(0.75)	(–1.68)	(1.55)	(1.29)	(–0.62)
Innovate	–0.492	0.547***		0.633	–4.430	0.363
	(–0.96)	(2.62)		(1.24)	(–1.06)	(0.72)
Age	0.066	0.027	0.275**	–0.252	2.080	0.270	–0.079
	(0.15)	(0.27)	(2.07)	(–0.95)	(0.69)	(0.60)	(–1.39)
Sales	0.413***	0.809***	0.773***	1.072***	3.700***	0.794***	0.770***
	(2.91)	(17.28)	(23.27)	(5.23)	(3.78)	(5.22)	(39.23)
Location	0.053	–0.247	0.174	1.213**	–9.700**	–0.681	0.404***
	(0.07)	(–1.53)	(0.58)	(2.28)	(–2.31)	(–1.09)	(4.66)
Export	0.540	–0.229	0.231	–0.374	–5.780	0.174**	0.232*
	(0.98)	(–1.10)	(0.55)	(–0.41)	(–0.97)	(2.05)	(1.69)
Country dummy							–0.382***
							(–3.50)
Constant	5.793***	–0.138	0.578	–3.629	–6.060***	–0.146***	1.569***
	(2.86)	(–0.19)	(1.04)	(–1.23)	(–3.43)	(–5.26)	(5.05)
Observations	90	435	1 643	42	247	279	2 125
R-squared	0.50	0.70	0.48	0.60	0.11	0.24	0.57
F()	35.57	155.48	107.99	7.28	4.12	12.54	345.28
P-value	0.0000	0.0000	0.0000	0.000	0.104	0.000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%. Export is a dummy variable equal to 1 if firm directly exports some of its products/services. Export Intensity is the percentage of sales gained from exports. Foreign is a dummy variable equal to 1 if more than 10% of the firm is owned by foreign individuals or entities. Innovate is a dummy variable equal to 1 if firm introduced a new product into the market or significantly improved a product/service in the last three years. Age is a firm's age and is calculated by subtracting the year the firm started operations from the survey year. Location is a dummy variable equal to 1 if the firm is located in the capital city. Obstacle measures the extent to which labour regulations are an obstacle to a firm's operations; it takes values from 1 to 5 (with 1 standing for no obstacle and 5 standing for very severe obstacle). Fire is the number of workers a firm would fire if it did not have to comply with labour regulations. Hire is the number of workers a firm would hire if it did not have to comply with labour regulations. Job reallocation is the summation of Fire and Hire. Investment is the amount spent purchasing machinery, vehicles, equipment, land and building in the previous year (a year prior to the survey year). Employment is the number of employees employed by the firm.

 Tables 9 and 10 show the regression results when the interaction terms ForeignInnovate and ForeignExport are included in the regressions. The variable ForeignInnovate was created by interacting the variables Foreign and Innovate. The variable ForeignExport was created by interacting the variables Foreign and Export. The variable ForeignInnovate was found to be insignificant across all sample countries (see Table 9). The coefficient of ForeignExport, however, was significant and positive for Kenya, Nigeria and Swaziland, suggesting that foreign-owned firms which export tend to employ more workers. The coefficient of ForeignExport was also positive and significant in the investment equation as shown in Table 10 (for Nigeria and Uganda). Concerning the relationship between ForeignInnovate and investment we find a positive relationship, for Swaziland and Uganda (see Table 10).

Table 9: Results for the impact of labour regulation on employment (including interaction terms)

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.006***	0.005***	0.037**	–0.026	0.039***	–0.284	0.008***
	(3.36)	(2.76)	(2.01)	(–0.59)	(3.34)	(–1.20)	(5.00)
Foreign	–0.026	–0.290	0.199	–0.513	0.056	0.169	0.458***
	(0.07)	(–1.54)	(1.12)	(–1.21)	(0.22)	(0.69)	(3.86)
Innovate	0.073	0.382***		–0.152	0.023	0.206**
	(0.25)	(4.09)		(–0.48)	(0.24)	(2.23)
Age	0.291*	0.118**	0.165***	–0.280	0.152**	0.278***	0.159***
	(1.92)	(2.05)	(4.14)	(–1.33)	(2.12)	(3.65)	(6.97)
Sales	0.352***	0.393***	0.329***	0.440***	0.381***	0.350***	0.311***
	(6.19)	(17.58)	(31.49)	(4.61)	(14.73)	(13.97)	(38.58)
Location	0.031	0.178**	0.039	–0.665*	0.056	0.008	–0.006
	(0.14)	(2.22)	(0.87)	(–1.80)	(0.53)	(0.07)	(–0.15)
ForeignInnovate	0.117	0.222		–0.221	0.190	–0.068
	(1.27)	(0.90)		(–0.42)	(0.69)	(–0.25)
ForeignExport	0.339	0.552**	0.727***	1.264***	0.304	0.335	0.120
	(1.30)	(2.45)	(4.12)	(3.22)	(1.13)	(1.56)	(0.84)
Country dummy^a							0.452***
							(9.99)
Constant	–2.62***	–3.852***	–3.302***	–1.454	–4.449***	–4.410***	–2.780***
	(–3.37)	(–11.56)	(–21.23)	(–0.99)	(–10.74)	(–9.69)	(–21.97)
Observations	90	578	1 643	51	246	280	3 985
R-squared	0.49	0.60	0.52	0.63	0.66	0.61	0.52
F()	31.50	115.48	459.70	13.73	78.55	53.42	436.29
P-value	0.000	0.000	0.0000	0.0000	0.0000	0.0000	0.000

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%. ForeignInnovate is the interaction of variables Foreign and Innovate. ForeignExport is the interaction of Foreign and Exports. ^aKenya is the reference country.

Table 10: Results for the impact of labour regulation on investment

Independent variable	Country
	Botswana	Kenya	Nigeria	Swaziland	Tanzania	Uganda	All countries
Job Reallocation	0.008***	–0.008	0.026	0.019	–1.030	–0.141	0.001
	(3.37)	(–1.41)	(0.91)	(0.33)	(–1.44)	(–0.11)	(0.61)
Foreign	–0.306	0.266	0.145	–2.707*	4.600***	–0.160	–0.169
	(–0.41)	(0.49)	(0.40)	(–1.96)	(4.33)	(–1.16)	(–0.83)
Innovate	–1.124	0.460**		–0.289	–1.040	0.227
	(–1.40)	(2.05)		(–0.45)	(–0.24)	(0.44)
Age	0.122	0.016	0.275**	–0.293	2.800	0.314	–0.071
	(0.27)	(0.16)	(2.08)	(–1.17)	(0.96)	(0.73)	(–1.25)
Sales	0.436***	0.802***	0.776***	1.082***	3.580***	0.708***	0.778***
	(2.89)	(18.40)	(23.64)	(6.37)	(3.86)	(5.02)	(41.53)
Location	0.204	–0.251	0.192	0.855	–1.080***	–0.623	0.400***
	(0.29)	(–1.55)	(0.65)	(1.49)	(–2.65)	(–1.03)	(4.55)
Foreign × Innovate	1.158	0.436		2.197*	–4.090***	0.114
	(1.15)	(0.84)		(1.98)	(–3.62**)	(0.74)
Foreign × Export	–0.273	–0.190	1.640***	0.060	–2.420	0.592***	0.250
	(–0.45)	(–0.40)	(4.64)	(0.05)	(2.38)	(4.86)	(0.92)
Country Dummy							–0.336***
							(–3.28)
Constant	5.756***	–0.017	0.534	–3.214	–6.340***	–0.129***	1.424***
	(2.78)	(–0.03)	(0.98)	(–1.27)	(–3.81)	(–5.06)	(4.84)
Observations	90	435	1 640	42	247	279	2 125
R-squared	0.40	0.70	0.48	0.66	0.17	0.28	0.52
F()	13.60	138.12	259.11	7.73	6.00	14.47	339.12

Notes: t-statistics are in parentheses. *Significance at 10%, **significance at 5%, and ***significance at 1%.

6. Conclusion and policy recommendations

The importance of labour market flexibility has received significant attention in the literature and in practice. A more flexible market can easily absorb shocks, and therefore can hire more workers or lay off workers when necessary. Proponents against such flexibility argue that a more flexible labour market is detrimental for the workers. Proponents of more labour market flexibility argue that a more flexible labour market results in more employment; while a more restrictive labour market, by making it more difficult for employers to fire, also makes it more difficult for employers to hire – because they become more reluctant to hire, fearing that it would be extremely difficult to get rid of the worker when the need arises in future.

We use a sample of about 4700 African firms to investigate the impact of restrictive labour regulation on a number of economic outcomes and find that more restrictive labour market regulations are detrimental to export propensity and export intensity. Although the results are sometimes mixed, in a majority of cases the results seem to suggest that restrictive labour regulations may be detrimental to employment and investment. Policy-makers must be cautious, however, when implementing employment regulations because too flexible regulations may benefit employers at the expense of employees, significantly reducing worker welfare. Highly flexible labour markets can also result in unemployment and/or underemployment as employers use more capital at the expense of labour. It can also result in increased casualisation of workers because employers may attempt to reduce costs by employing workers on short-term contracts rather than on a permanent basis.⁸

There is also evidence that a significant number of regulations in a number of African countries are largely ineffective and they do not seem to affect firm behaviour as intended. African governments must therefore emphasise both the crafting and implementation of economic policies. Where law must be enforced it should be judiciously executed so as to send accurate signals to all players in the economy.

The results from this study must be cautiously interpreted. First, variables such as export intensity and EPLs were not directly estimated. The proxies used to estimate these variables may affect the quality of the results. Second, the paper assumes that EPLs in place are efficiently enforced; in a number of African countries this may not be the case. Future studies can extend the study by using data that adequately capture the important variables used. Third, the data used are based on firm-level surveys conducted by the World Bank in many countries. Since the data are cross-sectional, the results may be affected by endogeneity.

Notes

³Flexibility can be understood as the capacity to respond rapidly and efficiently to changing economic environments.

⁴One should note that there are many other factors in addition to labour market flexibility that affect foreign direct investment, such as infrastructure, size of the market, human capital, and so forth.

⁵A labour market is inflexible if the level of unemployment insurance benefits is too high or their duration is too long, or if there are too many restrictions on the freedom of employers to fire and to hire, or if the permissible hours of work are too tightly regulated, or if excessively generous compensation for overtime work is mandated, or if trade unions have too much power to protect incumbent workers against competition and to control the flow of work at the site of production, or perhaps if statutory health and safety regulations are too stringent (Solow, 1998).

⁶The Tunisian revolution started in December 2010 and resulted in the fall of the Tunisian government led by President Ben Ali in January 2011.

⁷For a description of the World Bank firm-level surveys, see: http://www.enterprisesurveys.org/.

⁸Permanent contracts entail the payment of pension and medical aid expenses, expenses that employers can avoid paying if workers are employed on short-term contracts.

Appendix A. Variable description and definitions

Table A1: Variable description

Variable	Description
Export	Dummy variable: 1 if firm directly exports some of its products/services, 0 otherwise.
Export Intensity	The percentage of sales gained from exports
Foreign	Dummy variable: 1 if more than 10% of the firm is owned by foreign individuals or entities, 0 otherwise.
Innovate	Dummy variable: 1 if firm introduced a new product into the market or significantly improved a product/service in the last three years.
Age	This is a firm's age and is calculated by subtracting the year the firm started operations from the survey year.
Location	Dummy variable: 1 if the firm is located in the capital city, 0 otherwise.
Obstacle	This variable measures the extent to which labour regulations are to a firm's operations. It takes values from 1 to 5 (with 1 standing for no obstacle and 5 standing for very severe obstacle).
Job Reallocation	Fire + hire
Fire	Number of workers firms would fire if it did not have to comply with labour regulations.
Hire	Number of workers firms would hire if it did not have to comply with labour regulations.
Investment	Amount spent purchasing machinery, vehicles, equipment, land and building in the previous year (a year prior to the survey year).
Employment	The number of employees employed by the firm.