Conflict resolution processes,uncertainty and labour demand

Abstract

In this article, the impact of diminishing levels of uncertainty on labour demand, as a consequence of conflict resolution processes, is tested by means of a case study of a European region largely affected by political violence. For this purpose, the response of Basque manufacturing employment during conflict resolution attempts is used as a natural experiment with which to evaluate the effect of reduced uncertainty on labour demand. Accordingly, using the difference-in-differences technique, which overcomes some of the shortcomings of previous studies, the relative performance of Basque labour demand during the last two attempts to bring peace to the region is quantified. The longest ceasefire episode ever declared in the region is shown to have triggered a reactivation in labour demand and, therefore, that Basque manufacturing firms responded positively to the reduction in uncertainty by significantly raising their average number of employees. More precisely, it is found that the average number of workers employed by Basque manufacturing firms increased considerably when credible peace talks directed towards the end of the conflict were undertaken. Thus, when compared with their counterparts in similar Spanish provinces, the truce declared in 1998 boosted employment demand in Basque companies by more than 4%, which reflects the economic dividend of peace to be reaped in the event of an eventual conflict resolution and the establishment of a peaceful environment in the region.

Keywords

conflict resolution labour demand peace dividend uncertainty

Introduction

Since the seminal works of Barro (1991), Alesina et al. (1992) and Alesina & Perotti (1996) the existence of social unrest, terrorism and political instability have been documented as having adverse effects on the economies in which they take place. Thus, as new and expanded datasets have become available, scholars have devoted increasing attention to investigate both the consequences of political violence and the mechanisms that transform this violence into economic damage, and estimate the magnitude of their impact on diverse economic variables (see Frey, Luechinger & Stutzer, 2007; Sandler & Enders, 2008).

Even though it is commonly accepted that there is an inverse relationship between different sources of instability and economic outcomes, there has been little effort made to categorize and measure the importance of the various channels through which political conflicts hamper economic growth.¹

A political conflict that involves a permanent terrorist threat entails an increase in uncertainty in the economy and amplifies the inherent risk that business persons face when undertaking economic decisions. For that reason, a region subject to continuous conflict should manifest the distortion caused by increasing uncertainty in a significant redirection of economic resources away from productive uses and, therefore, reduced factor demands.

The Basque Country in Spain (henceforth, BC), which has been subject to a protracted and intense history of bloodshed, constitutes an appropriate case in which to evaluate the effects of periods of attenuated violence (i.e. reduced uncertainty) on labour demand. Terrorist activities have taken place in the region for more than 40 years and the continuous intimidation that the Basque society has endured is reflected by the distortion and depression that the economy endures (Abadie & Gardeazabal, 2003). Consequently, this article recapitulates the effects of the most recent ceasefires on Basque manufacturing companies’ labour demand.²

The results of the estimation show a substantial and positive impact of the 1998 ceasefire on the average size of Basque manufacturing firms’ labour force. Hence, when compared with their counterparts in comparable Spanish provinces, the truce declared in 1998 boosted labour demand in Basque companies by more than 4%, which shows the prospective benefits to be reaped from the eventual cessation of violence in the region.³ As a by-product, it can be seen that financial variables play an important role in determining labour demand at the firm level.

The remainder of this article is organized in the following way. The next section briefly reviews the theoretical framework. After that, I describe the sources and main characteristics of the variables available for the panel of Spanish manufacturing firms, allowing the empirical model of labour demand proposed to be estimated. Then, the estimates of the model are presented. Finally, I offer my concluding remarks.

Theoretical framework

In the works of Bernanke (1983), Caballero (1991) and Pindyck (1988, 1991), to mention a few, the irreversible investment literature emphasizes what can be gained by waiting out an uncertain environment. Not investing brings a cost in reduced profit streams, but it also brings a benefit in the form of an option to commit to taking on investments at a future date when the currently uncertain state of affairs has been alleviated. Thus, for example, Driver & Moreton (1991, 1992) reported that a higher level of uncertainty led to lower fixed investment in UK manufacturing.

The general idea behind the theory denominated ‘real options’ is that each investment project can be assimilated, in its nature, to the purchase of a financial call option, where the investor pays a premium price in order to obtain the right to buy an asset for a certain time at a predetermined price (exercise price), and possibly different from the spot market price of the asset (strike price).

The value of the option depends not only on the systematic risk of payoffs from the asset underlying the option, but also the variability of the asset’s payoffs (Dixit & Pindyck, 1994). Therefore, the total risk (both systematic and unsystematic) can be referred to as uncertainty in the ‘real options’ literature.

Similarly, the firm, in its investment decisions, pays a price (the cost of setting up the project) which gives the firm the right to use the capital (exercise price), now or in the future, in return for an asset worth a strike price. Taking this approach into account, the calculus of profitability for each single investment project has to consider the following three characteristics of the investment decision:

There is uncertainty concerning future rewards from the investment,

There is some leeway regarding the timing of the investment, and

The investment is partially or completely irreversible.

Recently, the real options theory has been applied to the case of employment and the determination of hours worked (see e.g. Chen & Funke, 2002) and, considering the at least partially irreversible nature of inputs of production, real options and irreversibility have been shown to play an important role in the short-run dynamics of investment and labour demand (Bloom, 2000).

Hence, Chen & Funke (2002) verify in their article that the theory of ‘real options’, when applied to the case of employment and working time determination, can be a fruitful extension of the traditional deterministic framework since it is able to consistently combine the existing interactions between irreversibility, uncertainty and the choice of timing, all peculiar characteristics of an employment decision.⁴

When there are fixed costs of either hiring (i.e. screening and training costs associated with the recruitment of a new employee) or firing (i.e. severance costs imposed by legislation when dismissing an employee), the firm will compare the option value of maintaining its current position against the alternative of hiring or firing.⁵

Accordingly, my a priori hypothesis hinges on the idea that a reduction of uncertainty caused by a ceasefire period would encourage firms to expand capacity and hire additional employees since they would expect the future to show an invigorated economic atmosphere in the region. Put another way, the option to hire is valuable if firms expect demand to increase in the future as a result of a less unsettling environment.

Furthermore, Bloom (2000) points out that one important effect of real options and irreversibility is to reduce the investment and hiring response to a demand shock. This occurs because firms will act more cautiously when capital and labour are partially irreversible and their market conditions are uncertain – any investment and hiring represents a risk from which the firm cannot easily extricate itself if conditions turn bad.

For that reason, I hypothesize that a positive and significant impact on employment demand in the BC should only occur when expectations of a future increase in demand are plausible enough, and this is likely to happen when a ceasefire becomes credible.⁶

Data description

In order to analyse the dynamics of employment demand at the firm level, I have collected data from SABI, the

Figure 1.

Average number of employees in manufacturing (1995 = 100)

Spanish and Portuguese version of Amadeus. The database is jointly elaborated by the Bureau Van Dijk and Informa, collecting the information from company accounts deposited in the Trade Registry Office of each country. Moreover, it refers exclusively to companies and excludes individual business persons.

Accordingly, this database comprises micro-data for approximately 600,000 businesses across the Spanish territory in any given year for the period stretching from 1995 to 2006. Regarding regional coverage, this is relatively uniform across Autonomous Regions, with employment coverage above 80% in Madrid, 60% in Catalonia and around 50% in the Basque Country for the year 2004.⁷ As a result, the sample of 54,366 manufacturing companies is unbalanced, with a ratio of about 6 to 1 compared with the comparison group.

In Figure 1, I represent graphically the evolution of the average number of employees in the manufacturing companies that comprise my sample of Spanish firms for the period stretching from 1995 to 2006.

At first glance, I observe that the progression in the number of employees in both groups is very similar for the whole period. As I will show later, this is a desirable characteristic of my choice of CM (the group formed by the firms located either in Catalonia or Madrid) as a suitable comparison set of companies.

Table I.

Average number of employees in manufacturing firms in Basque Country and comparison provinces

		1998 ceasefire
Group	Non-truce	1998–99	1998	2006
BC	33.4281	39.0790	41.4629	33.7532
BC	(141.7023)	(131.9366)	(137.5119)	(168.9836)
CM	33.4240	37.9210	37.3494	32.7774
CM	(261.3408)	(274.1396)	(232.9826)	(327.0564)

Source: Author’s calculations from SABI.

Standard errors in parentheses.

As mentioned above, the information supplied by companies to the Trade Registry Office – including data on the number of employees per firm – is most likely computed at the end of the year; therefore, this characteristic of the data shapes the observed evolution of the series.⁸ Accordingly, since the 1998 truce ended in November 1999, the corresponding figure for Basque companies could, to some extent, be taking into account the adjustment on the expectations of economic agents to a new scenario in the region.⁹

At this stage, I can proceed to the comparison of the average number of employees in manufacturing firms for the BC and the comparison provinces for the truce periods (i.e. 1998–99 or 2006) and the non-truce periods (i.e. 1995 through 2006, excluding the truce years) to see if, at a first glance, there are any important features worth mentioning. Additionally, I present the same information for the initial year of the 1998 ceasefire separately.¹⁰

Examining the information displayed in Table I, in the non-truce period, the average size of manufacturing firms in the BC does not differ systematically from that of the comparison provinces. Moreover, examining the period 1998–99, when the first of the ceasefires analysed was effective, it can be seen that the size of the average firm in both groups increased substantially compared with the non-truce period.¹¹

Nevertheless, it can be seen that in 1998 Basque firms augmented their size to more than 41 workers per firm – well above the average for the combined period – whereas CM companies had an average size below the mean for the joint period of truce (even though the difference in means is not statistically significant). This could be indicative of a significant and positive relative performance of Basque manufacturing employment in the initial year of the truce.

Finally, in the last column of Table I I present the average size of Basque and CM companies for the year 2006. On one hand, the number of employees in Basque manufacturing firms was above average for the non-truce period, but the increase is almost negligible, with an average number of workers per firm around 34. On the other hand, CM companies experienced a small decline in the size of their labour force with an average of one worker less than their Basque counterparts (the difference is not statistically significant).

Therefore, the observation of the data presented above suggests that firms in both groups of provinces increased the average size of their workforce during the truce announced in September 1998.¹² The average number of employees in Basque manufacturing firms during the truce was approximately 17% higher than it would have been otherwise, while the CM companies increased theirs by 13.5%.

As mentioned above, the database is not free of limitations and, for example, SABI only collects information on the number of employees per firm, which is the endogenous variable, but not on hours worked.¹³ An additional shortcoming of the dataset is the impossibility of identifying the type of workers in the firm, which prevents us from analysing whether the observed increase in the average size of Basque manufacturing companies was due to the inclusion of temporary or fixed-term workers, skilled or unskilled labour force, etc.

Empirical model of labour demand

In this section, I build an empirical model for a firm’s labour-demand decision, assuming implicitly the presence of adjustment costs to the desired level of employment and changing levels of demand uncertainty. Consequently, my working hypothesis, which is incorporated in the model specification through a dummy variable controlling for the effect of the truces on Basque employment, is that during ceasefire periods (i.e. periods of reduced uncertainty) firms located in the BC should increase their employment levels more than those of their counterparts in other Spanish regions and relative to periods of violence. Therefore, a more than proportional increase in the average number of workers in Basque firms with respect to others located in comparable regions and with respect to non-truce periods should be evident.

For that purpose, I adjust my empirical labour-demand model to account for the fact that Basque firms are supposed to react more positively than other Spanish businesses to expectations of peace following a ceasefire, introducing two ‘treatment effect’ terms as additional regressors referring to 1998 and 1999, which were the two years that the truce was effective.

In this way, my model resembles the ‘treatment effects’ models and, following the difference-in-differences approach, this allows us to account for the specific behavioural differences of Basque companies during a truce period (see Meyer, 1995; Bertrand, Duflo & Mullainathan, 2004). Thus, in the labour-demand equation, I include three additional covariates apart from the standard regressors that emerge in the literature: a dummy variable (bc) that takes the value 1 if the firm is located in any of the Basque provinces, and 0 otherwise; second, dummy variables for the years, (y98) and (y99), which control for time-period-specific shocks throughout the truce, common to all firms in the industry; and finally, two ‘treatment effects’ or interaction terms, (bcy98) and (bcy99), which denote the relative differential impact of the ceasefire on Basque firms’ average employment in 1998 and 1999, respectively.

To isolate the effect of ceasefire periods on Basque firms’ employment, I must control for all those variables or characteristics of the companies in the sample which make them behave differently when making hiring or firing decisions. For example, I have included as a covariate the costs of the labour input. For that purpose, I have calculated wage (wage) as the ratio of the total personnel expenses and the labour input employed at the firm.¹⁴

Furthermore, I approximate the firm’s technology with a group of observable firm characteristics – ; for example, the firm’s age (age), which controls for the way technology affected the company in the past, as well as for differences in efficiency.¹⁵ Moreover, I have to address several data issues arising from the microeconomic nature of the dataset, as well as account for firm heterogeneity. Consequently, I include firm-fixed effects in my estimation of Equation (1).

Additionally, I experiment with the inclusion of financial variables in my empirical labour-demand model to evaluate their influence on its determination. In this respect, following the analysis of Nickell & Nicolitsas (1999), I use the ratio of interest payments to cash flow (fin) in order to capture the financial pressure of the firm, picking up both the premium in borrowing costs and the probability that credit is constrained.

Since panel data are available, I use the following specification for the empirical labour-demand model for firm i and year t:

\begin{aligned} E M P_{i t} = & exp [α_{i 0} + α_{1} B C_{i t} + α_{2} T R U C E_{t} \\ + α_{3} B C_{i t} * T R U C E_{t} + α_{4} l n W A G E_{i t} \\ + α_{5} A G E_{i t} + α_{6} l n F I N_{i t}] + ϵ_{i t} \end{aligned}

Therefore, the key parameter in the estimation is α ₃, which accounts for the difference in the differences between the BC and the control regions (treatment effect). Put another way, the magnitude of the parameter on the interaction term gives us the effect of the truce on Basque average employment for manufacturing companies when compared with the ‘non-truce’ period and other Spanish provinces. A positive and significant α ₃ would therefore confirm a relative positive impact of the ceasefire on Basque labour demand.

In Equation (1) above I take the exponential of the right-hand side, since in this way I assure that the expectation of the endogenous variable always takes a positive value. Moreover, I have included an additive error term since it is equivalent to its multiplicative form. The key identifying assumption is that α ₃ would be 0 in the absence of the treatment, or E[∊_it | bc_it*truce_t ] = 0.

It is worth observing that I do not anticipate a permanent effect of a ceasefire on the average employment level of Basque companies – rather, a temporary one. This is because I am assuming that the main way a truce affects employment decisions is by the expectations with regard to higher demand in the future consequent of renewed hopes of a nonviolent scenario in the region, and these expectations can change very rapidly with good and bad news concerning the evolution of the conflict resolution process.

Now I turn to defining the optimal estimation strategy and choose the appropriate estimator for the labour demand Equation (1). To estimate the effect of periods of ceasefire on employment demand for Basque Country firms, I use what have come to be known as ‘count data models’ (Cameron & Trivedi, 1998). This type of model is particularly convenient when the interest variable is a count variable, as is the case at hand with the number of employees.

Accordingly, it is possible to obtain a consistent estimator for the conditional mean without specifying the conditional distribution. In fact, the Poisson regression model is able to do so even if the Poisson distribution is invalid. This is because the first-order conditions of the maximum likelihood problem are, generally speaking, more valid, so I can obtain a consistent estimator using the quasi-maximum likelihood approach, as discussed in Wooldridge (2001: section 19.2).

Results of the empirical analysis

I use Equation (1) to analyse whether a sustained period of truce stimulated employment demand in Basque companies. It should be noted that the effect of a ceasefire period on employment demand is estimated in relative terms with respect to the performance of businesses located in comparable Spanish provinces and the periods preceding and posterior to the ceasefire.

Owing to the estimation strategy – I estimate using conditional fixed-effects Poisson (quasi-maximum likelihood) estimation with clustered standard errors by firm – the normalization operation eliminates the effects of all those firm-specific characteristics which do not present time variation. Hence, the remaining explanatory variables are: the year effects for 1998 and 1999 (or 2006); the interaction terms for the same years and the Basque firms, which are the covariates of interest; and other control variables such as the age of the firm (age) or the personnel expenses per worker (wage).^16,17

The first two columns of Table II present the results of the analysis for the 1998 ceasefire considering a symmetric time frame around the truce – 1996 to 1997, and 2000 to 2001 – as a comparison period. As can be seen in the table, the interaction coefficients for the BC and the years in which the truce took place (i.e. bcy98 and bcy99) are both positive, but only the coefficient from 1998 is statistically significant. Therefore, the results confirm the assumption; a prolonged period of truce, which brought renewed hopes of peace and reduced the level of uncertainty in the region, encouraged Basque business persons to make hiring decisions, the consequence of which was an enlargement in the average employment level of Basque manufacturing firms.

Table II.

Labour demand estimates, 1998 ceasefire (individual year effects)

Dependent variable: employment by firm	1996–2001			1996–2000
Dependent variable: employment by firm	(1)	(2)	(3)	(4)	(5)	(6)
y98	–0.0148	–0.0065	–0.0067	–0.0141*	–0.0094	–0.0094
y98	(0.0108)	(0.0106)	(0.0107)	(0.0085)	(0.0091)	(0.0091)
y99	–0.0030	0.0054	0.0053
y99	(0.0064)	(0.0076)	(0.0076)
bc*y98	0.0380**	0.0408**	0.0405**	0.0350**	0.0345**	0.0344**
bc*y98	(0.0191)	(0.0197)	(0.0197)	(0.0148)	(0.0156)	(0.0156)
bc*y99	0.0105	0.0152	0.0151
bc*y99	(0.0196)	(0.0213)	(0.0214)
age	0.0668***	0.0744***	0.0745***	0.0717***	0.0797***	0.0797***
age	(0.0050)	(0.0051)	(0.0051)	(0.0049)	(0.0053)	(0.0053)
lwage	–0.5482***	–0.5741***	–0.5741***	–0.5800***	–0.5952***	–0.5952*** (0.0726)
lwage	(0.0521)	(0.0617)	(0.0617)	(0.0588)	(0.0726)	–0.5952*** (0.0726)
lfin		0.0228***	0.0193***		0.0202***	0.0196*** (0.0074)
lfin		(0.0057)	(0.0065)		(0.0060)	0.0196*** (0.0074)
lfin2			–0.0010			–0.0002 (0.0011)
lfin2			(0.0011)			–0.0002 (0.0011)
No. of firms	28,816	23,132	23,132	25,106	20,145	20,145
No. of observations	121,691	93,755	93,755	94,130	72,788	72,788
R-squared	0.022	0.025	0.025	0.023	0.025	0.025

Conditional fixed-effects Poisson (quasi-maximum likelihood) estimation; clustered standard errors in parentheses; standard errors for fixed-effect model clustered by firm.

***statistical significance at the 1% level; **statistical significance at the 5% level; *statistical significance at the 10% level.

To be precise, the results of the estimation reveal that during the largest truce that has ever taken place in the Basque territory, the average size of Basque manufacturing firms increased by approximately 4% with respect to their counterparts in comparable Spanish provinces and relative to the non-truce years.

Hence, it can be observed that the positive effect of diminishing levels of uncertainty occurred very rapidly and, therefore, the size of the Basque labour force employed in manufacturing firms increased significantly in 1998.¹⁸ Even though the positive effect of the truce can still be seen in 1999, it is, in statistical terms, non-significantly different from the figure in comparable regions.

All of the remaining coefficients have the expected sign. Hence, the age of the firm is positive and statistically significant and the natural logarithm of the average wage paid by firms is negative, indicating an inverse relationship between labour input and its remuneration.

Additionally, to ascertain whether the functional form of the model is correctly specified, I carried out a diagnostic Ramsey RESET test introducing squared fitted values (xb2) as an additional regressor in Equation (1). The estimated coefficient on xb2 was never significant at standard statistical significance levels, suggesting that the null hypothesis cannot be rejected, indicating that the model is correctly specified.

Several robustness tests to verify the consistency of the results have been implemented; for example, I have varied the length of the time frame adopted, estimated only with respect to the periods previous or posterior to the truce, or, in the fashion of Abadie & Gardeazabal (2003), used the case of Catalonia as a placebo study in order to verify the internal validity.¹⁹

In the second column of Table II I introduce the ratio of interest payments to cash flow in order to test the relevance of financial variables in the determination of labour demand at the firm level. The parameter accompanying lfin – the measure of financial pressure favoured by Nickell & Nicolitsas (1999) – designed to capture the premium on borrowing costs or the probability of credit being completely constrained, is statistically significant with a coefficient (robust standard error) of 0.0228 (0.0057), suggesting a relevant role for financial indicators in determining the level of employment.

Summarizing, the empirical estimates of the labour-demand functions reveal substantial adjustment in the average number of employees at the firm level in response to the ceasefire when differences in wages, financial pressure and other observable firm characteristics are controlled for.²⁰

Since it can be observed that the significant effect of the truce period took place in the initial year of the truce, in columns (4) to (6) of Table II I report the estimates of Equation (1), with and without the financial terms, considering exclusively the impact of the truce on the average level of employment of Basque manufacturing firms for the year 1998.

As can be seen, the results are consistent with those obtained previously and indicate a relative positive impact on labour demand for Basque manufacturing firms in the initial year of the truce. Hence, the average firm size of Basque companies in 1998 was 3.5% above the corresponding figure for the firms in the comparison group. The coefficient on the treatment effect is positive and statistically different from zero, indicating that the effect of the ceasefire was a relatively positive performance in the BC’s average number of employees. Likewise, the coefficient of the financial variable is also positive and significant: 0.0202 (0.0060).

Now I turn my attention to the second of the ceasefires I am able to analyse given the time coverage of the dataset. The truce was declared on 22 March 2006 and lasted for nine months. Thus, in Table III I report the outcomes of the estimation of Equation (1), where I have to adapt the specification to match the new truce period. Consequently, I have transformed the time dummies for 1998 and 1999 into a year dummy corresponding to 2006 (y2006), and their corresponding interaction terms with the BC into a term that considers the effect of the truce for Basque firms in 2006 (bcy2006).

Table III.

Labour demand estimates, 2006 ceasefire

Dependent variable: employment by firm	2004–06
Dependent variable: employment by firm	(1)	(2)	(3)
y06	–0.0156	–0.0143	–0.0136
y06	(0.0182)	(0.0193)	(0.0185)
bc*y06	–0.0050	–0.0272	–0.0271
bc*y06	(0.0230)	(0.0212)	(0.0214)
age	0.0598***	0.0732***	0.0727***
age	(0.0235)	(0.0264)	(0.0259)
lwage	–0.6722***	–0.8024***	–0.8015***
lwage	(0.0528)	(0.0625)	(0.0619)
lfin		0.0044	0.0125
lfin		(0.0054)	(0.0091)
lfin2			0.0021
lfin2			(0.0015)
No. of firms	32,649	22,638	22,638
No. of observations	89,886	60,248	60,248
R-squared	0.019	0.022	0.022

Conditional fixed-effects Poisson (quasi-maximum likelihood) estimation; clustered standard errors in parentheses; standard errors for fixed-effect model clustered by firm.

***statistical significance at the 1% level.

It should be observed that, due to data constraints, it was not possible to jointly compare the behaviour of firms during the truce with the previous and posterior periods, as I did earlier, and I had to estimate the effect of the 2006 ceasefire only in comparison to former years.²¹

The results show a negative, though non-significant, effect of the truce for Basque manufacturing employment in 2006. Hence, it seems that the truce that took place that year did not have any relative positive effect on the average size of manufacturing companies in the BC. This is a very interesting result since it leads the way to further research regarding the determinants of successful (i.e. in terms of fostering employment) ceasefires.²²

As mentioned above, my a priori hypothesis hinges on the idea that a positive impact on labour demand should only be observed when a credible ceasefire is declared. In this respect, the 2006 ceasefire is not regarded as a convincing event for several reasons, which supports the results obtained. First, the previous ceasefire (i.e. the 1998 truce) is considered by the vast majority to be a great loss of opportunity to achieve peace in the Basque Country, and that the terrorist group was responsible for its rupture. Second, the ceasefire was declared in a context of profound weakness for ETA, and therefore it was perceived as a trick by the terrorists to re-arm (e.g. during the truce, ETA stole 350 pistols in France). Moreover, the truce was declared in a phase of low violence levels – there had been no killings since May, 2003 – due to the intense pressure security forces put on terrorist activities; therefore, an attenuation of violence was not perceived. Third, Spanish society was expecting an announcement in which the terrorist group declared the abandonment of weapons and the end of the armed campaign, not a new ceasefire. Four, during the truce ETA did not stop their extortion campaign directed at Basque business persons. Finally, during the truce there were no significant advances in the peace talks, and from the beginning they were not supported by the principal opposition party.

Summary and conclusions

It has been argued that the ongoing political conflict and social unrest in the Basque Country has caused severe disruption in various aspects of life in this Spanish region. In the most influential article to date, Abadie & Gardeazabal (2003), using synthetic control methods, identified a comparison region and estimated that the Basque GDP per head was roughly 10% below the level that would have been achieved had the Basque Country not been subject to terrorism. Nevertheless, the determinants of this apparent decline in Basque income – and welfare – have not been completely identified, albeit some efforts have already been undertaken.

For example, Myro, Colino & Pérez (2004) singled out productive investment as one of the economic aspects more profoundly affected by the conflict and showed the importance of uncertainty in determining the pace of progress of investment in this Spanish region. In this article, I have focused on the other factor of production which could potentially be affected to a larger extent by a violent and uncertain environment – employment.

The magnitude and the statistical significance of the estimated parameters indicate that there is substantial adjustment of employment at the firm level in response to periods of renewed expectations regarding the end of the conflict. Thus, the employment response at the micro level to changes in the level of uncertainty about the future of the Basque Country is sizeable. Accordingly, in 1998 the average number of employees for Basque manufacturing firms was roughly 4% higher than the corresponding figure for firms in comparable regions and compared with the non-truce period.

On one hand, these results suggest that the negative economic performance observed in this Spanish region is reflected in low utilization of labour. On the other hand, the relative positive performance of labour demand during the largest and more credible ceasefire period ever declared in the region gives an idea of the potential benefits that the Basque economy would enjoy if a definite cessation of violence eventually were to occur.

This last fact has a special significance given the current state of affairs regarding the Basque conflict, since on the 10 January 2011 ETA declared a ‘general, permanent and verifiable’ ceasefire that could become the definitive one. Even though the previous two truces analysed in this article did not succeed in bringing peace to the region, I hope they paved the way for the ultimate end to the conflict.

Footnotes

Replication data

The dataset, codebook and do-files for the empirical analysis in this article can be found at . STATA 10 SE was used to obtain the estimates reported in the article.

Acknowledgements

I would like to express my deepest gratitude to the supervisor of my thesis project, Joao MC Santos Silva, for his valuable guidance and great experience that have helped me throughout this study. I have also greatly benefited from discussions with Michele Belot, Alejandro Cuñat, Marco Francesconi, Tim Hatton and Gordon Kemp. Additionally, I wish to place on record my appreciation of the input received from the participants at the RSS seminar at the University of Essex and the members of the International Economy Division at the Bank of Spain. Though I am the person solely responsible for the outcome of this article, I am in debt to its primary examiners, Alison Booth and Anja Shortland, for the discussions and advice received.

Notes

References

Abadie

Alberto

Gardeazabal

Javier

(2003) The economic costs of conflict: A case study of the Basque Country. American Economic Review 93(1): 113–132.

Alesina

Alberto

Perotti

Roberto

(1996) Income distribution, political instability and investment. European Economic Review 40(6): 1203–1228.

Alesina

Alberto

Ozler

Sule

Roubini

Nouriel

Swagel

Phillip

(1992) Political instability and economic growth. NBER working paper, No. 4173, September.

Barro

Robert J

(1991) Economic growth in a cross section of countries. Quarterly Journal of Economics 106(2): 407–443.

Bernanke

Ben S

(1983) Irreversibility, uncertainty and cyclical investment. Quarterly Journal of Economics 98(1): 85–106.

Bertrand

Marianne

Duflo

Esther

Mullainathan

Sendhil

(2004) How much should we trust differences-in-differences estimates? Quarterly Journal of Economics 119(1): 249–275.

Blomberg

Brock S

Hess

Gregory

Orphanides

Athanasios

(2004) The macroeconomic consequences of terrorism. Journal of Monetary Economics 51(5): 1007–1032.

Bloom

Nick

(2000) The real options effect of uncertainty on investment and labor demand. IFS working paper 00/15, November.

Booth

Alison

Chen

Yu-Fu

Zoega

Gylfi

(2002) Hiring and firing: A tale of two thresholds. Journal of Labor Economics 20(2): 217–248.

10.

Caballero

Ricardo J

(1991) On the sign of the investment– uncertainty relationship. American Economic Review 81(1): 279–288.

11.

Cameron, A

Colin

Trivedi

Pravin K

(1998) Regression Analysis of Count Data. Cambridge: Cambridge University Press.

12.

Chen

Yu-Fu

Funke

Michael

(2002) Working time and employment under uncertainty. CESifo working paper 732(4), May.

13.

Colino

Alberto

(2012) Conflict resolution processes, uncertainty and investment dynamics: Evidence for the Basque Country. Defence and Peace Economics, iFirst Article: 1–17, DOI: 10.1080/10242694.2012.673840.

14.

Dixit

Avinash K

Pindyck

Robert S

(1994) Investment under Uncertainty. Princeton, NJ: Princeton University Press.

15.

Driver

Ciaran

Moreton

David

(1991) The influence of uncertainty on UK investment. Economic Journal 101(409): 1452–1459.

16.

Driver

Ciaran

Moreton

David

(1992) Investment, Expectations and Uncertainty. Oxford: Basil Blackwell.

17.

Fielding

David

(2003) Modelling political instability and economic performance: Israeli investment during the intifada. Economica 70(277): 159–186.

18.

Frey

Bruno S

Luechinger

Simon

Stutzer

Alois

(2007) Calculating tragedy: Assessing the costs of terrorism. Journal of Economic Surveys 21(1): 1–24.

19.

Gaibulloev

Khusrav

Sandler

Todd

(2008) Growth consequences of terrorism in Western Europe. Kyklos 61: 411–424.

20.

Jametti

Mario

Brülhart

Marius

Schmidheiny

Kurt

(2007) Do agglomeration economies reduce the sensitivity of firm location to tax differentials? York University Working Papers 9.

21.

Meyer

Bruce

(1995) Natural and quasi-natural experiments in economics. Journal of Business and Economic Statistics 12(1): 151–162.

22.

Myro

Rafael

Colino

Alberto

Pérez

Patricio

(2004) Crecimiento económico y conflicto social en el País Vasco [Economic growth and social conflict in the Basque Country]. In: Buesa

Mikel

(ed.) Economía de la secession. El proyecto nacionalista y el País Vasco [The economics of secession: The nationalist project and the Basque Country]. Madrid: Ministerio de Hacienda, IEF, 47–74.

23.

Nickell

Stephen

Nicolitsas

Daphne

(1999) How does financial pressure affect firms? European Economic Review 43(8): 1435–1456.

24.

Walter

(1962) Labor as a quasi fixed factor. Journal of Political Economy 70: 538–555.

25.

Pindyck

Robert S

(1988) Irreversible investment, capacity choice and the value of the firm. American Economic Review 78(5): 969–985.

26.

Pindyck

Robert S

(1991) Irreversibility, uncertainty, and investment. Journal of Economic Literature 29(3): 1110–1148.

27.

Sandler

Todd

Enders

Walter

(2008) Economic consequences of terrorism in developed and developing countries: An overview. In: Keefer

Philip

Loayza

Norman

(eds) Terrorism, Economic Development, and Political Openness. Cambridge: Cambridge University Press, 17–47.

28.

Wooldridge

Jeffrey M

(2001) Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB