I The Idea of a Civil Service
We can think of the civil service simply as the people employed in the public sector, with the task of providing public services. This corresponds to the everyday use of the term. The public administration literature, however, is much more restrictive: it requires both independence and meritocracy. Independence implies that the civil service consists of career bureaucrats who advise their political masters and faithfully execute their programmes, irrespective of whether or not they share their views. In return, they are secure in their position: they cannot be fired for their political views and remain in position when the government changes. Meritocracy means that the hiring, firing, and career advancement of civil servants are determined solely by their professional competence and performance, not by any payments they make or by their religion, political allegiance, ethnic affiliation, or social class.Footnote 4 The meritocratic civil servant in this model is incorruptible. While he must be paid well, so as not to be led into temptation, his main motivation is the esteem (and self-esteem) acquired through his contribution to the public interest.
It is worth stressing that such a civil service is not a Western invention. For example, imperial China had already instituted a system of selecting civil servants on the basis of merit some 1,400 years ago: henceforth, prospective mandarins had to pass a famously difficult entrance examination. Once they had been admitted to the civil service, their independence was encouraged by not stationing them in their region of origin and by frequently transferring them to another part of the empire during their career. When Marco Polo described the mandarin system, it had already functioned in China for some 700 years, while at that time in Europe the public sector was still basically an unprofessional extension of the court, serving at the whim of the prince and his entourage.
In Europe, there was no independent and meritocratic civil service until 1800. In Great Britain it dates from the late nineteenth century. Pressure to replace patronage by selection on the basis of merit had already mounted in the 1850s when the disastrous British experience in the Crimean War revealed that the civil service was almost as chaotic and incompetent as the military.Footnote 5 The resulting public outcry led to the famous Northcote–Trevelyan report of 1854, which proposed drastic reforms. However, no action was taken for a very long time. Only in 1870 did Gladstone as prime minister adopt its recommendations for an independent, politically neutral civil service, selected on the basis of merit and paid well enough to resist corruption. Most other Western countries adopted similar innovations.
But notable exceptions remain to this day. In the United States, political appointments are still quite common: an incoming president is entitled to replace some 4,000 civil servants. One result is that many American ambassadors are millionaires who have distinguished themselves with campaign donations rather than as career diplomats.
Independence has an obvious implication for any study of the impact of the civil service: consequences (good or bad) of policies cannot be attributed to the civil service that has implemented them. For example, if policy choices made by politicians result in economic growth or stagnation, then that in itself says nothing about the quality of the civil service. It is after all possible that bad policies are implemented by a high-quality civil service, or indeed vice versa.Footnote 6 Similarly, if educational outcomes are very poor, as they are in Tanzania, then this may be due to policies (e.g. bad teacher training programmes, low salaries) or to their implementation by civil servants (e.g. teacher absenteeism).
This represents, of course, a problem for empirical analysis. There now exists a booming literature (which uses experimental methods) on how the functioning of the civil service can be improved. This provides evidence on, for example, what incentives can reduce teacher absenteeism. But there is no established methodology for establishing the past impact of the civil service; any attempt to do so faces the daunting task of disentangling the effect of the policies themselves from the way they were implemented.
In many African countries, the civil service tradition left by the colonial powers quickly crumbled after independence. That system was hardly meritocratic to begin with (since it involved racial discrimination) and a purely meritocratic system would have been very difficult to set up at the time: educational attainments were still quite low. (In 1962, Tanganyika – as it then was – had only a handful of people with Masters degrees; President Nyerere was one of them.) More importantly, in many countries the civil service was seen, not primarily as the provider of public services, but as the key channel for patronage. The ruling group used its control over the state to provide benefits to its supporters, notably in the form of public sector jobs, which provided income, security, and prestige. The provision of public services was therefore in itself of limited importance; it mattered mainly to the extent that it guaranteed or threatened the survival of the regime.
An important implication of the use of the civil service as a tool for patronage was that real wages in the public sector were fixed far in excess of market-clearing levels. This induced massive migration from rural to urban areas.Footnote 7 Excess supply in the urban labour market required mechanisms for allocating the scarce jobs. The mechanisms that were adopted typically favoured particular regional or ethnic groups.
Patronage was not the only reason why a civil service in the strict sense, that is independent and meritocratic, disappeared or failed to emerge in many African countries. Under many regimes, notably those we would now call developmental states (Adam and O’Connell, 1999). Weberian independence was not considered necessary or desirable: political leaders and their civil servants were seen as jointly involved in nation building; a strict separation of their roles (choosing policies and implementing them) would have been considered counterproductive.
II Tanzania 1962–1982
Tanzania’s experience in this period was exceptional. After having gained its independence in December 1961, it initially followed a familiar path: indigenisation of the civil service combined with continued heavy reliance on expatriates who wrote all the key planning documents. The Arusha Declaration marked the end of that period. Tanzania became exceptional in Africa in at least two ways: by embarking on a far-reaching attempt, unique in Africa, to build a socialist economy,Footnote 8 and by adopting strict policies designed to keep civil servants honest.
In Tanzania, the key economic policies of that period – import controls, production in parastatals, price controls, communal farming, and pan-territorial agricultural pricing – led to a balance of payments crisis and economic stagnation. This was temporarily veiled when Tanzania benefited from a huge term of trade improvement, the beverages boom of 1975–9, but it became clear in the early 1980s when exports collapsed, serious shortages emerged, and people reverted to barter trade (Bevan et al., Reference Baynes1990). It would, of course, be absurd to attribute this disastrous outcome to the quality of the civil service. In fact, in my view, prior to the collapse, the Tanzanian civil service did a remarkable job of carrying out an almost impossible task.
Notably, Tanzania had in the 1970s one of the least corrupt civil services in Africa, in spite of the huge opportunities for corruption created by the system of controls. In the 1960s, Nyerere had already introduced the ‘leadership code’, a set of rules that prohibited middle- and senior-ranking party members and civil servants from holding shares in, or being a director of, a private company; renting out houses; receiving more than one salary; or employing others. Pratt comments:
He did not see the leadership rules as the imposition of a morality which was not accepted. They were the enforcement of an ethic that was still part of the values of society. It was an ethic that was being rapidly replaced in the towns, amongst the leaders and within the civil service by an aggressive acquisitive individualism.
Nyerere saw clearly that if people in powerful positions were allowed to maximise their own income, then his vision of a socialist economy could not be realised. He and his advisers, such as Reginald Green and Justinian Rweyemamu, both Harvard PhDs, failed to see, however, that the system of controls erected in the 1960s and 1970s created enormous incentives for ‘acquisitive individualism’, for example in the form of the very large gap that emerged between the official and the black-market exchange rate. The control system therefore helped to undermine the very ethic that Nyerere hoped to enforce.
This chapter on the Tanzania civil service addresses three questions:
How much has the civil service contributed to economic development?
What constrained its performance?
How can it be reformed so as to maximise its contribution to economic development?
I will consider these three questions in turn.
III The Contribution of the Civil Service to Economic Development
Regrettably, how much the civil service has contributed to economic development is not an empirical question that can be answered rigorously. The reason is that it is impossible to build a convincing counterfactual: one cannot make comparisons over time (did outcomes differ between periods with or without a civil service?) or across space (did outcomes differ between areas with and without a civil service?).Footnote 9 One can, of course, determine whether changes in the size or the functioning of the civil service were followed by changes in outcomes. Strictly speaking, this establishes, of course, only correlation, not causation. Interpreting the outcomes as causal effects would amount to the post hoc ergo propter hoc fallacy, since economic outcomes have many determinants other than the quality of the civil service. In fact, an ‘omitted variable’ may well be responsible for both a change in the civil service and a change in economic outcomes. Concluding that the one causes the other would then clearly be a mistake.
This is a common and fundamental problem in economic history. As usual, the best one can hope to do to explain economic outcomes at the national level is to arrive at a plausible account of causal effects.Footnote 10
Nevertheless, Mukandala does arrive at conclusions regarding causal effects. He notes, ‘Its support role in policymaking was minimal. The party was supreme and all major policy decisions of the time were made by the party’s NEC, which had a powerful secretariat.’ The party (TANU, later CCM) was powerful both in the design of policies and in the supervision of their implementation. In fact, but this is not in the chapter, there was next to the civil service a powerful parallel party structure that reached from the top down to the very lowest levels. (Under the ten-cell system there was a party official for every ten households, much as in China. The ten-cell leader conveyed party instructions to the households and reported on their activities.) Mukandala writes: ‘Since the civil service was not a lone or even the key player in economic development, it can only assume limited responsibility’ for the poor success rate of a long series of policies initiated in the Nyerere period. He concludes, and I agree, that the ‘grand policy decisions’ then taken were detrimental to economic growth. Mukandala adds:
Most of these policies were beyond the control of the civil service since the NEC of TANU had self-appointed itself as a supreme decision-making organ.
Here I disagree, for three reasons. First, that policy decisions were taken by politicians should not be seen as an aberration, but rather as the normal state of affairs.Footnote 11 That the civil service was excluded from the policymaking process tells us little on its contribution to development. At best, the exclusion of professional advice may have contributed to poor policy choices.
Secondly, that the party also played a major role in what is properly the arena for a civil service, policy implementation, is a major confounder for the analysis of the civil service’s contribution. Mukandala mentions it only in passing. In my view, while assessing the impact of a civil service rigorously would be extremely difficult in the best of circumstances, TANU’s role in policy implementation makes such an analysis impossible: there simply is no independent civil service in this period, so, for better or worse, outcomes cannot be attributed with any confidence to the way the service functioned.
Thirdly, in line with this, the chapter first seems to argue that the outcomes were very bad but cannot be attributed to the civil service. However, the chapter also argues:
Generally, the findings indicate that, since the late years of Ujamaa, Tanzania’s civil service has been characterised by work misbehaviour such as shirking of responsibilities (absenteeism and negligence) and corruption and misappropriation of resources (e.g. through procurement and creation of ghost workers), and generally it has failed to regulate opportunistic politicians and even colluded with them, at the expense of the economy.
You cannot have your cake and eat it too: either the contribution of the civil service was ‘minimal’, and it had only ‘limited responsibility’ for what went wrong, or ‘it has failed to regulate opportunistic politicians and even colluded with them, at the expense of the economy’. By seemingly trying to have it both ways, the chapter leaves unclear what its answer is to the first of the three questions.
IV Constraints on the Performance of the Civil Service
What constrains the civil service depends of course on what it is supposed to do: a lack of qualified candidates, for example, is a constraint for a Gladstonian civil service but not for a service that is meant to provide patronage.
Within the Gladstonian framework, there are three issues to consider. First, what are the incentives for people to join (and remain in) the civil service? Incentives can include pay, security, prestige, and opportunities for corruption. The issue is whether the package offered by the civil service (officially or not, including the opportunities for engaging in bribery) is attractive compared with an alternative, for example employment in the private sector. In other words: can the civil service attract and retain high-quality candidates? Secondly, is the civil service’s independence guaranteed or is its functioning hampered by political interference? Thirdly, are hiring, promotion, and firing based exclusively on merit, or do non-meritocratic criteria such as ethnicity or party membership play a role?Footnote 12
Regarding the first issue, there are several indications in the chapter that incentives are indeed problematic. Mukandala shows, for example, that there are (and have been) major staff shortages, notably in health and education. He explains that these reflect a freeze on public sector employment, but he also refers to ‘those who left for greener pastures’.Footnote 13 In what way these other pastures were greener is not discussed. This is a pity since an analysis of the incentives for joining or leaving the civil service must be the basis for any proposed reform.
Table 5.3 presents very interesting evidence from a recent World Bank study, comparing quality indicators in the public and private sector. Perhaps surprisingly there appear to be no large differences. If the public and private sectors do not really differ in terms of workload or absenteeism, then what makes the civil service relatively attractive or unattractive, and how has this changed over time? Real wages in Tanzania fell dramatically from the late 1960s until the mid-1980s; they rose again in the mid-1990s. Unfortunately, there is no evidence in the chapter on changes over time in the relative attractiveness of the two sectors, public and private.Footnote 14 Apparently, the situation in 2015 was such that public wages were ‘outcompeting those in the private sector’ (Mukandala is quoting Leyaro et al., 2014, p. 10). However, the chapter stresses that two major problems remain: intrinsic motivation has drastically declined since the days of Nyerere, and, while incentives are quite attractive in the higher ranks, motivation is quite low for those in lower ranks, the majority of civil servants. Corruption has become a major issue, and the chapter makes it plain that this has a long history, going back to the late Ujamaa years.
On the other two issues, the chapter is very clear. In the 1960s and 1970s, politicisation severely undermined the civil service’s neutrality: it could no longer freely use its professional expertise. The party had no time for the setting of priorities or for moderation. Since that time, the civil service has become more independent. However, the relationship between civil servants and politicians remains linked by patronage; in that sense, the civil service is not completely independent.
There was a period when all university graduates were recruited by the public sector and patronage (in the form of hiring friends and relations) was a major issue. However, from 1999 there have been meritocratic reforms, including the introduction of open competition and non-discrimination. Forged or questionable academic certificates used to be accepted, but have now become the basis for a large number of dismissals.Footnote 15 The discussion of these reforms is tantalising but very brief. I would have liked to know more about how they were designed and, most importantly, how they survived in spite of the fact that they must have run into opposition from vested interests. Did Tanzania find ways to overcome this problem that are relevant for future reforms or for reforms in other countries?
The recent changes seem to be in the direction of greater independence and meritocracy. The chapter makes abundantly clear that there still is a long way to go. For example, Mukandala describes that candidates from different institutions must be treated equally in spite of large quality differences and wryly notes ‘the results have been disastrous’. Moreover, connections still matter, most promotions are still automatic, and party membership can still trump qualifications. There is clearly scope for deeper reforms.
V Reforming the Civil Service
Recall that the third question asked how the civil service can be reformed so as to maximise its contribution to economic development. The chapter suggests numerous improvements: better recruitment, rigorous competition, merit-based evaluation, scrapping unnecessary posts, continuous checking, an improved incentive structure with pay tied to verifiable performance criteria, stronger accountability, action against misbehaviour, better time management, and improved ICT use. It is hard to disagree with this list.
Reform must, of course, be selective: an attempt to address all these issues simultaneously would be a recipe for failure. The standard approach to setting priorities in civil service reform is largely technocratic: it focuses on the low-hanging fruit and postpones reforms that are costly or aimed at less urgent problems. Essentially, this amounts to ordering reform measures in terms of cost–benefit ratios.
Mukandala correctly notes that this approach is unlikely to lead to sustainable reform. Measures intended to reduce political influence on the civil service or aimed at eliminating the scope for patronage will be resisted by politicians, who will quickly recognise that they undermine their position. This is, of course, the central issue in any political economy analysis of reform, whether it be of the civil service, trade policy, decentralisation, taxation, or industrial policy. In all such cases, the challenge is to design reforms that will not be blocked by a set of agents in a position to do so, notably, but not necessarily, those in power. Essentially, the ruling class must perceive a reform package as an improvement of their position in spite of the fact that the reforms themselves directly threaten their position. Typically, this will require some form of compensation.
This issue of making the unacceptable acceptable is at the heart of the political economy of reform. I note, not as a criticism but with some regret, that the chapter does not address this. Given the importance of this issue, this makes the discussion similar to Hamlet without the Prince of Denmark. I had hoped to learn what the key sensitivities were that would generate opposition to reform and in what way that resistance might be overcome. That hope was disappointed, but I recognise, of course, that these are extremely difficult questions to answer, even in a speculative way.
Open access