Institutional legacies and HRM: similarities and differences in HRM practices in Portugal and Mozambique

Abstract This is a study of institutional change and continuity, comparing the trajectories followed by Mozambique and its formal colonial power Portugal in HRM, based on two surveys of firm level practices. The colonial power sought to extend the institutions of the metropole in the closing years of its rule, and despite all the adjustments and shocks that have accompanied Mozambique’s post-independence years, the country continues to retain institutional features and associated practices from the past. This suggests that there is a post-colonial impact on human resource management. The implications for HRM theory are that ambitious attempts at institutional substitution may have less dramatic effects than is commonly assumed. Indeed, we encountered remarkable similarities between the two countries in HRM practices, implying that features of supposedly fluid or less mature institutional frameworks (whether in Africa or the Mediterranean world) may be sustained for protracted periods of time, pressures to reform notwithstanding. This highlights the complexities of continuities which transcend formal rules; as post-colonial theories alert us, informal conventions and embedded discourse may result in the persistence of informal power and subordination, despite political and legal changes.

countries. Institutional approaches to comparative HRM suggest that both formal institutions (such as regulatory and economic systems and the roles of employment-related bodies such as the role of trade unions and collective bargaining) and informal institutions (embedded shared understandings of desirable ways of doing things) impact on, and are moulded by, firm-level practices. If path dependence means that the ex-colonial country maintains many of the features and practices of its colonial past, attempts at changing institutions may have less dramatic effects than is commonly assumed. Whilst recent theoretical work on Africa's economic progress suggests that colonial powers in contexts of large European settlements devoted more care to building better institutions that proved durable following independence (Acemoglu & Robinson, 2012), Mozambique, despite a very large settler community, suffered both economic collapse and instability in the early years of independence. We explore the reasons behind this from a firm-centred perspective, and drawing on post-colonial theory, explore why changes in formal regulation had only a limited effect on actual workplace practices.
The broad literature on comparative HRM evaluates the linkages between national contexts and predominant modes of people management, and seeks to compare these between settings (Brewster, Mayrhofer, & Morley, 2004). Initially, the main focus was on the advanced societies, but by the early 2000s, there was a growing interest in emerging markets. HRM in Africa remains relatively under-researched (Ellis, Nyuur, & Debrah, 2015;Kamoche, Debrah, Horwitz, & Muuka, 2004;Wood & Brewster, 2007; see also Horwitz & Budhwar, 2015). There is a very little research on HRM in Mozambique (Dibben, 2010;Dibben & Nadin, 2011;Sartorius, Merino, & Carmichael, 2012), and more generally, there remains a paucity of empirical comparative analysis and longitudinal studies of past and contemporary HRM and employment relations between former colonial regimes and their independent state offspring. This paper adds to the limited body of research by comparing employment relations and HRM in Mozambique with the former colonial power, Portugal. It also seeks to supplement existing cultural approaches (see Jackson, 2015), with a stronger focus on long historical processes. Comparisons between the analysis of the ex-colonial power and its former territory could shed significant light on present HRM and employment relations systems. The paper also engages with socio-economic institutional approaches to the analysis of work and employment in emerging economies (see, for example, Dibben & Williams, 2012;Wood, Dibben, Stride, & Webster, 2010;Wood & Frynas, 2006). Through examining employment relations in a colonial power (Portugal) and in its former colony (Mozambique), it seeks to understand the valence of arguments that point towards enduring colonial legacies vs. those that indicate the effects of the distinct differences in context. In doing so, it links post-colonial studies to comparative HRM and employment relations, and develops a deeper understanding of institutional analysis through its examination of an emerging economy.
The paper proceeds as follows: we first examine general issues of post-colonialism, formative institutional legacies, and Africa's economic performance, focusing on work and employment in Portugal and Mozambique. We then use this lens to compare the two countries and develop hypotheses, explaining how we have used two separate surveys to test these hypotheses. We present the findings, and then in the discussion and concluding section of the paper examine the contribution to post-colonial theory and comparative HRM and the implications for institutional theory.

Post-colonialism
Post-colonial studies are often dated from Said's (1978Said's ( /1985 volume. However, post-colonialism has changed its meaning over time from a strictly temporal one to a 'fighting' one (Lazarus, 2004, p. 4). Said's (1978Said's ( /1985 fundamental point was essentially that a 'scientific' discourse has emerged involving a set of values and assumptions about the role of Western countries and their colonial territories that is based on assumptions, many of which are probably wrong. Modern post-colonialism tends to be more nuanced, critiquing both imperialist and nationalist narratives (Rao, 2012). The process of neo-colonialism is arguably now not carried out by the former imperial states, but mainly by their multinational enterprises (MNEs) and the international organizations that work to ensure free trade (Petras & Veltmeyer, 2001). Having 'finally' emerged in management and organization studies (Jack, Westwood, Srinivas, & Sardar, 2011, p. 275), post-colonialism has been used to explain the markets of MNEs (Banerjee, Chio, & Mir, 2009) and the way in which organizations are structured and operate (Prasad, 2012). However, there are few studies of HRM in post-colonial societies that address this contested theory (McLeod, 2007) and even fewer that address the legacy of colonial institutions on HRM.
Most post-colonial studies focus on the discourse used to ensure the continued domination of the colonial masters, but there have been calls to take greater account of materialist formulations Murphy & Zhu, 2012), an approach that we adopt here, in linking institutional and post-colonial theories. We acknowledge that the process of transferring knowledge across national borders is itself imbued with colonialism (Frenkel & Shenhav, 2006), but we try to set our analysis within the political-economic framework of colonialism and its aftermath (Jack et al., 2011). Whilst engaging with post-colonial debates, we use comparative survey evidence to shed new light on the nature of long-term continuities in workplace practice, formative institutional legacies and institutional design.

Institutions and Africa's economic performance
Within the comparative institutional literature, there has been growing interest in seeking to explain the uneven and relatively poor and volatile performance of many African economies (Acemoglu & Robinson, 2012;Robinson, 2002;Wood & Horwitz, 2015). In an influential account, Acemoglu and Robinson (2012) argue that the relative fortunes of African nations can be traced back to the colonial era, and differences in colonial policy. In countries unsuitable for significant European settlement, there was a focus on constructing exploitative institutions, designed to extract wealth for the benefit of the colonial power. In countries with significant European settlements, more attention was accorded to promoting the rule of law, and replicating the institutional arrangements of the home country, providing a more stable basis for continuity and growth after independence.
Portugal was the poorest of the European colonial powers in Africa, but in lasting for more than 500 years, Portuguese colonialism in Africa endured far longer than that of any other major colonial power -Britain, for example, ruled most of its African colonies for less than a century. In late colonial era Mozambique, there was a very sizable European settlement, and attempts to extend the institutions of the metropole to the colony. Despite this, the country experienced much political and economic instability in the two decades after independence, making the country an important exception to this rule; the reasons for this are explored in this paper.

Institutions, work and employment in Portugal
The early literature on comparative capitalisms assumed that markets would gradually move to one or other of the liberal/coordinated market models, because of their greater efficiencies and complementarities (Hall & Soskice, 2001). Later accounts recognized that Mediterranean or mixed-market economies represented an archetype in their own right and that, despite apparent institutional changes, they continued to follow distinct trajectories (Amable, 2003;Hancké, Rhodes, & Thatcher, 2007), characterized by divisions between a historically statist sector (the public sector, state-owned enterprises and large private firms with close ties to the state) and a poorly regulated SME and informal sector. They had weaker and less closely coupled institutions than the more mature archetypes (Amable, 2003). Portugal has a history of institutional layering, whereby new institutional arrangements do not totally substitute for old ones (Rocha & Araujo, 2007), and rigid hierarchies and procedurality are accompanied by modernization (Tavares & Alves, 2006, p. 584). Continuities include the relative importance of agriculture, a large SME sector (Karamessini, 2008), and a relatively large informal economy, whilst informal networks facilitate the bypassing of formal institutional arrangements (Aguiar, 2005).
From 1933, authoritarian corporatist rule under Salazar's Estado Novo built on the Portuguese legal tradition and patriarchal dimensions of Portuguese culture (Aguiar, 2005). In the 1960s, the state moved towards greater integration with the European economy and, in its closing years, there were cautious moves towards liberalization (Nunes, Mata, & Valério, 1989). The economy internationalized, with a significant influx of foreign capital and further industrialization (Holman, 2001, p. 59). However, high infantile mortality, low educational levels and an inefficient agricultural sector remained, notwithstanding the large gold reserves that were accumulated during this period (Blakanoff, 1996).
In 1986, Portugal joined the European Union (EU), and the Portuguese Government initiated a process of liberalization and privatization, which would turn the country into one of the three largest privatizing countries in the Western world (Cunha & Cooper, 2002). The boom in Portuguese industry from the 1980s until the early 1990s was followed by a gradual decline in the face of European competition (Karamessini, 2008), and Portugal was marked by rising income inequalities (da Silva Lopes, 2003). Meanwhile, the family continued to play a central role in many Portuguese firms, with corresponding patriarchal, authoritarian and gendered divisions of labour and the persistence of family-based welfare and informal systems of HRM (Karamessini, 2008). Portugal became committed to ever greater labour market flexibility which built on a tradition of poor enforcement of labour law, with light oversight of small firms and often weak and ineffective workplace representation (Karamessini, 2008).
Considerable dispute and labour turmoil followed the fall of the Estado Novo (da Silva Lopes, 2003;Karamessini, 2008). This was ameliorated through tripartite dialogue, initiated in 1985. Under tripartite arrangements, the government, employers' confederations and trade union confederations formed part of the Permanent Committee of Social dialogue whose main role was to promote social dialogue on public employment policies, vocational training, social security, tax and public administration or fixing the amount of the minimum monthly wage. Thematic task forces, composed by the social partners, analysed relevant issues from a technical point of view before their final assessment. This process eventually led to a series of pacts surrounding collective bargaining, health and safety and vocational training and frequent government interventions, particularly in terms of minimum wages and guidelines for wage increases (Cardoso, 2006), but the state remained ineffective in enforcing emerging labour standards in the large informal sector.

Institutional legacies, work and employment relations in Mozambique
The Portuguese arrived in Mozambique in 1498. Most of the 500 years of Portuguese rule were characterized by high levels of exploitation of resources similar to those in West Africa, but the closing years of colonial rule were also associated with the large-scale transfer of people from the metropole (Hanlon, 1991). The exploitative nature of early Portuguese rule reflected the underdeveloped nature of Portugal itself. Early Portuguese colonial rule was mostly exercised via the prazos, concessionaires acting as feudal warlords (Newitt, 1969) and, in rural areas, peasants were subject to indirect rule via appointed regulos or chiefs. This period did not result in serious institution building beyond extending the state's monopoly over the use of armed force, and slowly modernizing administration within the cities. By the twentieth century, a large portion of Mozambique was administered by the 'majestatic companies' , created because the Portuguese state lacked the necessary human, financial and administrative resources to administer the colony directly. A coercive labour system was employed (Carvalho, 2012).
In 1961, the Estado Novo responded to growing pressures for decolonization by adopting a policy of integration, bolstered by greater investment in the colonies, more migration from the metropolitan homeland, and active efforts to develop industry (Henriksen, 1978). The late 1960s saw significant increases in inward foreign investment and industrialization, associated with Portuguese economic development plans (Planos de Fomento) for the colonies and the increased importance of the services sector (Fleshman, 2009;Francisco, 2003). Elements of the authoritarian corporatism of Portugal penetrated Mozambique, encompassing both employer interests and the white-dominated trade union branches of the Portuguese unions. However, late Portuguese colonialism failed to incorporate the African majority economically (Munslow, 1984). One consequence was the emergence of Frelimo, a political and military liberation movement that opposed the colonial and the capitalist system (Francisco, 2003).
Colonial era authoritarian corporatism was abandoned on independence in 1975. Frelimo formed the first post-independence government and embraced state socialism (Hanlon, 1996;Munslow, 1984). The racial elites lost their privileges and fled. Most enterprises and land holdings were placed under state ownership. However, private SMEs and entrepreneurs remained. In 1976, an armed rebel movement, Renamo, sponsored by white minority governments in South Africa and what is now Zimbabwe, emerged, and the country descended into a protracted civil war. In 1994, a political settlement led to Mozambique's first democratic elections, which were won by Frelimo. Yet, the government was forced to turn to the World Bank and the IMF and, again on a broadly similar timescale to Portugal, implemented sweeping neo-liberal reforms (Brück, 1998). In HRM terms, this contributed towards large-scale job losses (Dibben, 2010); whilst democratization meant more autonomy for trade unions, job losses resulted in a collapse in membership (Dibben, 2010). A new tripartite negotiating forum, the Consultative Labour Commission, set a minimum wage, but enforcement of agreements was uneven or weak (Webster, Wood, & Brookes, 2006). Throughout all these changes, there were strong continuities in Mozambican work and employment relations, with divisions between larger firms (often either currently or previously state owned), SMEs, and a very large informal sector, with a legacy of authoritarian management (Dibben & Nadin, 2011;Webster & Wood, 2005;Wood et al., 2010).

Methods
As outlined in the introduction, through examining employment relations in a colonial power (Portugal) and in its former colony (Mozambique), the key research problem tackled by this paper is how to understand the validity of arguments that point towards enduring colonial legacies vs. those that indicate the effects of the distinct differences in context.

Hypothesis development
In the context of these two countries' economic and institutional histories, to what extent are work practices in Mozambique likely to be similar to those in Portugal? Both Mozambique and Portugal experienced turbulent decades but had relative political and social stability, and well-developed labour legislation. Informal networks of support played an important role in the SME sector, given the failures of formal institutions and difficulties in gaining access to capital. In addition, both countries have a history of patriarchal authority systems at the workplace, a residue of a specific form of industrialization. However, there also appear to be a number of marked differences. Mozambique's primary institutional legacies have been extractive, in contrast to the uneven, but more modern, institutional tradition of Portugal. Hence, despite late colonial institutional building, including extending the coverage of mother country legislation into the colony, and the importance of informal networks and conventions, there are likely to be persistent differences in institutional capabilities, with the Mozambican environment being more precarious for both firms and workers. Both countries have undergone major neo-liberal reforms, but these were more intense in Mozambique. Given these differences, and in particular, the more precarious state of the Mozambican economy and weaker institutional coverage, job security is likely to be more fragile than in Portugal. Hypothesis 1a: Whilst the incidence of redundancies is likely to be relatively high in both contexts, they will be significantly higher in Mozambique.
Hypothesis 1b: The proportion of the workforce on non-permanent contracts will be high in both countries, but relatively higher in Mozambique.
Owing to common institutional legacies and shortfalls, both countries have experienced persistent skills gaps. The commitment of firms to training and development can be evidenced by its formality or informality (Goergen, Brewster, Wood, & Wilkinson, 2012). However, the lower capabilities of Mozambican organizations and the precarious nature of their competitiveness is likely to mean that training and development will be less developed than in Portugal.
Hypothesis 2: Informal workplace-based forms of training will be used in both countries, but will be more common in Mozambique than Portugal.
Given the importance of informal networks and obligations in both contexts, it is likely that informal methods of recruitment will be common in both cases but, given infrastructural limitations and general institutional weaknesses, their use might be more pronounced in Mozambique.
Hypothesis 3: The use of informal methods of recruitment will be more common in Mozambique than Portugal.
Collective voice mechanisms can involve participation by workers and their collectives in the re-negotiation of the employment contract, and in the setting of work relations including financial participation and communicative mechanisms (Marchington, 2005;Whitley, 1999, p. 39). Unionization and collective bargaining in both countries are likely to be strongly concentrated in larger firms but, given that Mozambique represents a more extreme case of organizational segmentation and incomplete institutional coverage, its use will be more pronounced in that country.
Hypothesis 4: Collective bargaining and unionization are likely to be concentrated in large firms in both the countries under review, but more so in Mozambique than Portugal.
Other employee voice mechanisms may be weak in both contexts. However, the tradition of patriarchal managerialism might suggest that general meetings, whereby managers inform workers of organizational developments in their own terms may be the most common communication device (Webster & Wood, 2005). Again, it could be argued that their use will be more pronounced in Mozambique than Portugal.
Hypothesis 5: Workforce meetings represent the principle form of communication in both contexts, but their use is more common in Mozambique than Portugal.
Although performance-related pay may be viewed as a component of hardline approaches to HRM (Gollan, Poutsma, & Veersma, 2006), in the developing world many organizations will simply opt for paying the lowest rate that the external labour market will bear. In such circumstances, piece work, a simple form of performance related pay, represents an alternative. However, it is possible that firms which opt to pay the bare minimum, reining in shirking through close surveillance, may still seize short-term competitive advantages.
Hypothesis 6: Performance related pay is used by a larger proportion of firms in Portugal than in Mozambique.

Data and samples
The data used for the empirical analysis within this paper come from two sources. For Mozambique, the data come from a survey of firms carried out in 2003, the sampling frame being compiled from the official telephone directory. The survey instrument was broadly based on the more established Cranet survey (see below), but shortened and adapted in order to focus on issues and concerns that might be present in a developing country context. In total, 177 firms were surveyed. Stratified sampling was employed, strata being defined by geographic locale and, within each geographic locale, by sector. This ensured representation of both key areas of industrial concentration and sector. The survey covered greater Maputo and greater Beira, the two principal industrial and commercial centres in Mozambique. For Portugal, the data employed in this paper are from the repeating Cranet survey which collects evidence on HRM issues of private and public organizations (Brewster et al., 2004). Given that the Mozambican survey aimed to broadly replicate a selection of the themes covered in the Cranet survey, this comparison is considered to be valid. The data-set used in this paper contains results from the 1999-2000 survey (a relatively close wave of the Cranet data to the 2003 Mozambique survey, and well in advance of the 2008 economic crisis) which is broadly representative with respect to the industrial sector in each individual country. Mozambique has gone through a number of quite dramatic, transitions, from colonial rule to state socialism, civil war and political uncertainty, to democratization and neo-liberalism. At the time that the survey was conducted, the latter changes had been bedded down and have persisted to this present day. As this study is on continuity in institutions and HR practice, despite external shocks, this makes the findings of the surveys still relevant; indeed, it could be argued that Mozambique's somewhat smoother progress since then (see, for example, World Bank Group, 2012), means that are there fewer pressures on firms to fundamentally change -or at least question -what they do. Information was collected from a postal survey of the senior HRM practitioner in organizations with more than 100 employees. Hence, the data available were from 177 firms in Mozambique and 169 firms in Portugal. Once listwise deletion was undertaken for the key variables in the empirical model, this was reduced to a working sample of 300 firms, with 157 in Portugal and 143 in Mozambique.

Methods of analysis
Although the data-sets are different and the questionnaires used in each country were not the same there are broad areas of consistency across the two. In terms of the formal hypotheses, there were relevant direct questions in both surveys. For example, all firms were asked if they had made use of compulsory and voluntary redundancies in the last three years and whether temporary or fixed-term contracts were used. They were also asked directly whether they made use of word of mouth to recruit as well as whether they used informal work-based training, thus enabling the construction of binary variables. The only exception was the coverage of percentages of trade union membership, which was measured as a continuous variable.
In order to test whether significant differences were present across the two countries, as stated in the hypotheses, it was important to distinguish between inter-country differences and those differences which resulted from factors below the country level, such as organizational size or the industry in which the organization operated. Therefore, empirical models were estimated. The key factors such as size, sector and industry were included as control variables (Porter, 1980;Hall & Weiss, 1967).
The variables representing each of the dimensions outlined in the hypotheses above were used as the dependent variable of a model. In addition, the variables were estimated as a function of size of organization, sector and type of industry. For trade union membership, the model was estimated using ordinary least squares and was of the form: with y i being the dependent variable, α the intercept term, u i a normally distributed error term, x i the vector of explanatory variables and β′ their estimated coefficients. In situations where the dependent variable was binary, a logit model was used such that: where P i is the probability that the dependent variable equals 1, 1 − P i is the probability of it being zero and L i is the log of the odds ratio. Since the log of the odds ratio was linear in the parameters, the logit model could be estimated in the linear form (Gujarati, 1995, pp. 554-556). When the various OLS and logit models had been estimated they were then re-estimated separately for the two different countries to enable testing of the stated hypotheses. For the OLS models, we used a version of the Chow test, where and RSS is the residual sum of squares from the pooled model, the Portugal model and the Mozambique model, respectively, n 1 is the number of observations from the Portuguese sub-sample and n 2 the same from the Mozambique sub-sample, finally k is the number of parameters. The F statistic followed an F distribution with degrees of freedom (k, n 1 + n 2 − 2k) (Gujarati, 1995, pp. 262-264).
For the logit models, we used a likelihood ratio test of the form: where LR is the likelihood ratio and LL is the maximization of the log-likelihood function from the Portuguese, Mozambique and pooled models, respectively. LR followed a chi-squared distribution with k degrees of freedom, where k was the number of estimated parameters (Greene, 2000, pp. 825-826). Both the OLS and logit models had the null hypothesis that the structure of the two sub-sample models was the same, i.e. for a given size, sector and industry Portuguese organizations tended to be the same as those in Mozambique in terms of that particular dimension.

Findings
Table 1 records the mean values by country for each of the key dimensions used in the analysis. For example, 51% of the organizations in Portugal had made use of voluntary redundancies in the previous three years compared to only 12% of those in Mozambique. Considering these figures in the context of the hypotheses enables an understanding of how the findings can be interpreted, at least at an observational level. The data enabled a separation between voluntary and compulsory redundancies and although more firms in Mozambique had made use of compulsory redundancies, around four times as many Portuguese firms had used voluntary redundancies. Therefore, the overall use was considerably higher in Portugal. Hence, Hypothesis 1a is rejected.
Hypothesis 1b is also not supported. However, our measures are fairly blunt since the question only asked if the organization made use of temporary or fixed-term contracts; the respondent was not asked about the proportion of staff employed under these contracts. Clearly, this issue is of potential interest and would be an obvious target for future research.
Informal work-based training and informal recruitment were both more common in Mozambique than in Portugal. For both informal training and informal recruitment the test null can be rejected, for the former at the 1% level and the latter at the 5% level. Although the levels of confidence differ, it is fairly safe to conclude that Hypotheses 2 and 3 can therefore both be accepted.
Hypothesis 4 was that collective bargaining and unionization was likely to be concentrated in larger firms in both the countries under review, but more so in Mozambique than Portugal. Collective bargaining was higher in Portugal, and union membership higher in Mozambique. However, the hypothesis posited a relationship between level (of collective bargaining and membership) and firm size (higher levels in larger firms). In fact, in the regression results upon which the structural tests are based, the size variable was not significant in any of the models. Therefore, although we can be fairly confident that there are significant differences within these dimensions across the two countries these are not influenced by firm size. Therefore, Hypothesis 4 is rejected.
Hypothesis 5 was that workforce meetings represented the principal form of communication in both contexts, but their use was more common in Mozambique than Portugal. The analysis showed that the use of workforce meetings was considerably higher in Portugal. We cannot tell whether such meetings were the organization's major communication channel with its employees. Finally Hypothesis 6 posited that performance related pay was used by a larger proportion of firms in Portugal than in Mozambique. In fact, it was used by a higher proportion of firms in Mozambique, although only marginally. Hypothesis 6 is rejected.
To understand more clearly whether the hypotheses outlined earlier can be accepted or rejected, we need to go beyond these observational level data to the structural tests described above. Table 2 reports the outcomes from carrying out these tests. In the interest of space and brevity, in addition to reflecting the fact that the explanatory variables were used as controls only (with no formal hypotheses related to them), the regression and logit results are not reported here, although they are available from the authors on request. Table 2 reports, for each of the dependent variables, the number of parameters in the model which established the degrees of freedom for the resultant test. It also reports the critical values at the 5 and 1% levels. In most cases, these values are based on the chi-squared distribution, the only exception being union membership where, because it is a continuous variable and an OLS regression was estimated, the critical values are based on an F (16,268) distribution. In each case, the null hypothesis was that there were no significant differences for that variable between Portugal and Mozambique. In order to accept the hypotheses outlined in this paper, the estimated statistic from the models needs to be higher than the critical value, and the null of the structural test rejected.
In summary, for hypothesis 1a, we cannot reject the test null for differences in compulsory redundancies, but we can for voluntary redundancy. So overall, hypothesis 1a is rejected. For hypothesis 1b (differences in the use of non-permanent contracts), the null test was strongly rejected, so hypothesis 1b is accepted. For informal training and informal recruitment, in both cases the null test can be rejected, for the former at the 1% level and the latter at the 5% level. Although the levels of confidence differ it is fairly safe to conclude that hypotheses 2 and 3 can both be accepted. Hypothesis 4 relates to the coverage of collective bargaining and the extent of union membership. Although in both cases the test null is rejected, both are more likely in larger firms. However, in the regression results upon which the structural tests are based, the size variable was not significant in any of the models. Therefore, although we can be fairly confident that there were significant differences within these dimensions across the two countries these were not influenced by firm size. Therefore, hypothesis 4 is rejected. Finally, both hypotheses 5 and 6 can be rejected. With the former, although the test null is rejected, workforce meetings were actually more commonplace in Portugal. With regard to Hypothesis 6, the hypothesis is rejected since there were no significant differences in the probability of organizations using performance related pay in each of the two countries.

Discussion and conclusions: the limitations of institutional embeddedness
In terms of HR practices and employment relations our results have shown that there were both similarities and differences between the two countries. In both countries under review, firms were similarly likely to make use of redundancies. At the time of the surveys, both economies had undergone long transitions to marketization, associated with previously large organizations shedding substantial amounts of labour which, in turn, had been unevenly mopped up by smaller firms and, particularly in Mozambique, by the informal sector. In Portugal, the major privatization programme in the 1990s led to corporate restructuring and downsizing, with workforce reduction primarily through voluntary redundancies, enabling a relatively peaceful labour relations climate (Cunha, 2000). Although we found a significant difference between the two countries in the relative propensity to make use of temporary labour, this was not on the lines predicted. Labour legislation is widely disregarded by many smaller and medium-sized employers in both countries, but state capabilities in Mozambique are weaker and, hence, enforcement less likely. If it is easy to shed labour, then the temporary contracts offer relatively limited advantages to employers, perhaps explaining why they were less commonly utilized in Mozambique than Portugal.
With regard to the differences in recruitment, training and communication, informality is greater in Mozambique. With regard to training, in Mozambique at the time of independence, 93% of the population was illiterate. Although great strides were made in the first years of independence, progress ground to a halt owing to the civil war (UNESCO, 2000). The radical structural adjustment policies in the 1990s further damaged state provision of education and technical training (Hanlon, 1996). These circumstances meant that firms were particularly likely to fall back on informal training. Informal recruitment methods were also more common in Mozambique, which might again reflect limitations in literacy and infrastructural weaknesses, or might be due to the operation of extended informal networks of support that interpenetrate many African countries (Hyden, 1983), and/ or the stronger emphasis on communitarian ties (c.f. Sartorius et al., 2012). An area where both countries had common ground was in relation to collective bargaining, which was concentrated amongst larger employers. Whilst this would be true of almost all economies, this segmentation appears to be particularly pronounced in the two countries under review, compared to firms in mature economies. This emphasizes the more marginal nature, and weaker capabilities, of medium and smaller firms operating in such contexts.
Whilst firms made greater use of informal 'on the job' approaches in Mozambique, in both countries there was an extensive reliance on general workforce meetings. Such forms of verbal, top-down, communication highlight limitations in employee voice: employees expressing views unpopular with management in such forums could open themselves to victimization. It also reflects a common culture of informality in managing people in both contexts.
The use of performance-related pay was similarly present in both countries, but such systems can be very diverse, ranging from sophisticated approaches to the use of piecework. However, in contexts where employment relations are widely individualistic, state enforcement of the law uneven and employee countervailing power weak, it can be argued that many firms simply default to paying the bare minimum that the market will bear.
Our evidence would have been stronger with more closely matched samples and with larger numbers of respondents. Despite these limitations, we believe that our findings are clear: in key areas, there are fewer differences than expected in people management and employment relations between Portugal and Mozambique, in contrast to other research using the Cranet survey which has found considerable differences between countries (see, for example, Goergen et al., 2012).
Our findings contribute to literature on post-colonialism, since they suggest that there are abiding institutional legacies from colonialism. As post-colonial theory alerts us, existing power relations and associated patterns of behaviour may be embedded and replicated through a process of discourse that transcends formal structures; this further explains the durable influence of colonial era informal conventions and norms, despite the wholesale departure of colonizers and the ideological bankruptcy of the colonial Portuguese state. Whilst looking to state socialist economic models, the post-independence Mozambican creole elite retained close cultural -and, indeed, at times, familial ties -to the metropole, making the resumption of trade and investment relations between Portugal and Mozambique in the 1990s smoother and facilitating the return of significant numbers of former settlers. This imparted not only continuity in managerial practice, but also promoted the revitalization of specific ways of doing things, even if this made workers no better or worse off. In other words, the persistence of aspects of colonial era formal and informal regulation and associated firm-level behaviour cannot be explained solely through an analysis of formal institutions.
At the same time, there were some limitations to the durability of institutions. In this respect, this study contributes to debates on the long-term influence of colonial level institutional arrangements on the fundamental fortunes of nations. Acemoglu and Robinson (2012) suggest that in countries with large settler populations, colonial authorities devoted more time to institution building and, more specifically, directly replicating the institutional arrangements of the metropole. They further suggest that this would result in greater political stability, societal coherence and economic progress (Acemoglu & Robinson, 2012;Wood & Horwitz, 2015). Yet Mozambique showed limited economic progress in the years after independence. The limitations in the durability of institutions may be accounted for by three possible, but not mutually exclusive explanations. The first is that Portugal's experiment at institution building under the Estado Novo was different to that adopted by other colonial powers since it was more centralized, with the aim of extending parent country institutions to draw in the colonies as provinces, rather than replicating them, as was the case, for example, with the British East and Southern African colonies in the late colonial period. Therefore, embedded rules lacked centralized bureaucratic support. Secondly, Portugal itself represented a relatively underdeveloped country with not wholly functional institutional arrangements; extending them abroad would not necessarily provide a sound foundation for growth. Thirdly, independence led to the exit of skilled colonialists, which challenged training institutions, leading to a greater emphasis on improvisation and the informal dissemination of skills. All of this suggests that institutional substitution or replication does not necessarily provide an adequate foundation for effective HRM, or, indeed, for firms to succeed in the same manner as in the metropole. Proponents of the liberal market model and associated firm-level practices -based around numerical and pay flexibility -might therefore do well to consider the range of challenges associated with the dissemination of a particular model of rules and practices into a very different context. More generally, a feature of liberalizing economies is that firms compete increasingly on cost grounds, with a tendency towards low value-added work and employment policies, processes that are particularly pronounced in peripheral countries (Tickell & Peck, 1992, p. 355). Although the economic crisis that began in 2008 has changed things since then, at the beginning of the decade pressures towards the adoption of neo-liberal policies in Portugal were limited by accession to, and membership of, the EU (Haworth & Hughes, 2003, p. 670). Hence, the consequences of Portugal's semi-peripheral geographic locale were ameliorated. At the same time, Mozambique's peripheralization has been accentuated. It is possible that the more recent scale of neo-liberal reforms in Portugal, in response to the economic crisis, may result in greater similarities with Mozambique.
Clearly, despite great differences in some areas, we found much common ground between these two geographically distant countries, external shocks and neo-liberal reforms notwithstanding. In terms of these findings' relevance for institutional theory, on the one hand, it could be argued that the common ground reflects abiding institutional legacies, with the modernization of institutions and economic life that took place in late colonial Mozambique leading to convergence with the key dimensions of the Portuguese model. As Pitcher (2002) notes, key features of Mozambican governance and society proved quite resilient despite independence, civil war and later neo-liberal reforms; it is as yet uncertain as to whether the 2008 economic crisis will have similarly limited effects on Portuguese institutions, although initial evidence suggests that marketization has, once more, gone hand-in-hand with structural continuities (Carneiro, Portugal, & Varejão, 2014). On the other hand, our findings may reflect practices that are relatively common amongst a broad range of firms operating within evolving institutional contexts. Hall and Soskice (2001) suggested that effective complementarities are likely to be concentrated within mature liberal and coordinated markets; other types of capitalism may be associated with less effective regulation and fewer advanced and mutually supporting intra-organization practices, leading to strong pressures of convergence to one or other of the two fully developed capitalist archetypes (Hancké et al., 2007). Later work has sought to identify a greater range of capitalist archetypes, including the 'segmented business systems' in Africa (Wood & Frynas, 2006) and 'informally dominated market economies' (Dibben & Williams, 2012). These archetypes are characterized by segmentation between larger organizations and their more weakly regulated smaller and medium-sized counterparts, ongoing (sometimes radical) liberalization, uneven state enforcement capabilities and shortfalls in needed skills and capabilities. Rather than adding to an already extensive list of possible varieties of capitalism, future research might aim to provide a more integrated taxonomy of fluid and partially mature institutional contexts. This would recognize that contexts cannot be dismissed as universally dysfunctional and note that their distinctive features may be sustained for protracted periods of time (rather than converging with one or other of the two mature varieties of capitalism). Operating within such contexts may be associated with relatively durable practices which, whilst sometimes low value added, may continue to confer advantages specific to the context in which they are encountered, even if simply in terms of imparting greater resilience in coping with external shocks.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the United States Department of Labour (USDOL) and the Federal Mediation and Conciliation Services (FMCS) who funded the research and provided assistance during the research phase of the Mozambican part of this study.