Gender,Education and Socio-economic Diversity in the Civil Service of Bangladesh: The Case of Administration Cadre

Abstract

This study explores the dynamics of the bureaucracy of Bangladesh, focusing on three main dimensions of diversity (demographic, informational and socio-economic) over 20 years from 2001 to 2020. In particular, the study investigates whether and how the distribution of administration cadre officials, in terms of gender, education and socio-economic status, has changed over this period and whether the pay scale of 2015 has affected the composition of officials working in the administration cadre service. In the case of gender (demographic) diversity, we found that the representation of female officials has gradually increased over time but showed a slight downward trend in the later periods. We have also found that since the introduction of the 8th pay scale, the bureaucracy has become more diversified regarding the officials’ educational (informational) background. In particular, the share of civil servants with an Engineering and Computer Science background has sharply increased during the 35th–37th BCS. In contrast, the share of officials with an Arts and Social Science background has dropped significantly. By analysing the education and occupational status of officials’ fathers, we observe that introducing the pay scale positively affects the diversity in bureaucracy by increasing the representation of officials from less well-off socio-economic backgrounds. Although the share of officials whose fathers were farmers showed an initial downward trend, it has marginally improved after introducing the pay scale. In a survey of admin cadre officials, we observed that the introduction of the pay scale had not affected the representation of officials from rural areas.

Keywords

Gender diversity socio-economic diversity representative bureaucracy rural representation

Introduction

Earlier studies on the bureaucracy of Bangladesh, especially studies conducted in the 1980s, 1990s and early 2000s, were highly critical of the government machinery. These studies painted the organ as an elitist, non-responsive one that did not try to address the needs and concerns of the marginalised and vulnerable groups of the society (Huque, 1997; Huque & Rahman, 2003; Jahan, 2006; Zafarullah, 1987; Zafarullah & Huque, 2001). Recent studies, however, indicate a different trend and show that the bureaucracy is becoming more responsive and citizen-friendly and is keen to come up with innovative solutions to assist the underserved groups of the population (Baniamin et al., 2020; Shahan et al., 2021).

Although these studies have indicated that the inclusion of a new generation of bureaucrats who are willing to represent and reflect the demands of the citizens at large is possibly changing Bangladesh’s service delivery spectrum, no detailed empirical study has been conducted to explore whether the demographic and socio-economic composition of the civil service has changed over time (Shahan, 2021).

Empirical studies indicate that passive representation (PR) can lead to active representation and will likely make the bureaucracy more responsive. As mentioned earlier, in Bangladesh, we have no idea about the bureaucracy’s ‘representativeness’. Moreover, we do not know whether (and to what extent) critical events like introducing a new pay scale in 2015 have affected PR. This research addresses this gap while relying on the ‘representative bureaucracy’ framework.

The primary objective of the research is to explore the dynamics of Bangladesh’s bureaucracy, focusing on the representativeness of different socio-economic classes over 20 years from 2001 to 2020. In particular, the study investigates whether and how the distribution of administrative officials, in terms of gender, education and socio-economic status, has changed over this period and whether the pay scale of 2015 has affected the composition of officials working in the administration cadre service.

Since the ‘greater representativeness’ of the bureaucracy is likely to result in ‘greater access to service delivery’ for the underserved groups, it is essential to understand how representative the bureaucracy of Bangladesh has become over time. Therefore, the central research question that the study is trying to address is—does the bureaucracy of Bangladesh mirror the demographic composition of the country, especially in terms of socio-economic characteristics, ethnic and gender composition, and educational attainments of the students who are eligible to apply for the civil service? Furthermore, it is important to understand that Bangladesh’s bureaucracy has undergone several changes over the last decade, and probably the most important of these changes is the introduction of a new pay scale—the National Pay Scale of 2015. In this study, we also want to understand the effect of introducing this pay scale on the representativeness of the bureaucracy. Therefore, our second research question is: What is the impact of the pay scale of 2015 on the representativeness of the Bangladeshi bureaucracy?

In this study, the representativeness of different groups in the bureaucratic services has been explored by investigating the level and trend in the diversity of administrative officials. A diversified workforce provides a favourable environment for developing innovative ideas. Studies have also shown that increased diversity in the bureaucracy leads to improved integrity in organisations. Choi et al. (2018) found that increased female representation in bureaucracy leads to increased organisational integrity. According to this study, an increase in organisational integrity may result from female bureaucrats being more likely to fight gender-based inequality than their male counterparts.

Moreover, the nature of diversity influences the creative activity in teams. Parthasarathy et al. (2011) identified three forms of diversity in work groups: demographic, informational and value diversity. Diversity in terms of age, gender and ethnicity is defined as demographic diversity. Differences among workforce members in terms of educational background are considered informational diversity. Finally, diversity in terms of values, beliefs and attitudes is categorised as value diversity. In addition to these three dimensions of representativeness, the diversity of bureaucracy in terms of socio-economic classes is a crucial indicator of PR in bureaucracy. Therefore, this study focuses on the administration cadre’s demographic, informational and socio-economic diversity. In particular, the study explores how diversity in bureaucratic services has evolved from 2001 to 2020 and how the introduction of the 8th pay scale has affected the diversity in civil service.

In the next section, we discuss the literature on representative bureaucracy, focusing on the evolution of the concept and its importance for delivering public services to citizens of the country. The third section presents the study’s methodology and explains the data used for empirical analysis. The study’s findings are analysed in the fourth section. Finally, the last section concludes the article.

Literature Review

The Evolution of the Concept

The idea of representative bureaucracy was first coined by Kingsley (1944), whose main argument was that if the government reflects the composition of its population in terms of skills, beliefs and class, it would be successful in protecting democratic values and representing its citizens. Kingsley considered bureaucracy as an instrument for establishing a democratic state, and as such, he was arguing for a bureaucracy that would reflect the marginalized classes of the society. Levitan (1946) and Long (1952) argued that since bureaucracy includes appointed unelected officials, the most effective way to ensure bureaucratic responsibility and accountability is to design an entity that reflects the population it serves.

Even though earlier studies on representative bureaucracy mainly highlighted its demographic composition, they did not explain how an increased civil service representativeness would positively affect the democratic state. Mosher made this critical contribution in the 1970s. Mosher (1968) noted that bureaucrats’ socialisation experiences play a significant role in developing their beliefs and values, influencing how they interpret their roles and responsibilities when entering the civil services. Therefore, the bureaucracy that represents the diverse socio-economic classes of society will be more responsive to the needs of different communities.

From this perspective, Mosher draws a connection between passive and active representation. Of these two, PR focuses on exploring whether the bureaucracy of a country mirrors the demographic origins of the population (of that country), especially in terms of race, gender, ethnicity, social class, or other characteristics. If PR is ensured in bureaucracy, it is expected to ensure inclusivity and democracy (Bowling et al., 2006; LeRoux, 2009; Moldovan, 2016).

Over the last 30 years, several studies have been conducted in different parts of the world, and these studies provide empirical evidence that PR is translated into active representation. In other words, previous studies found that if the bureaucracy mirrors the socio-economic features (such as average income and occupation of bureaucrats’ parents) and demographic characteristics of the country (measured in terms of gender, race and ethnicity), it would be more sympathetic, friendly and responsive to the needs of its citizens and more considerate in delivering services to the marginalised population (Keiser et al., 2002; Kennedy, 2014; Riccucci & Van Ryzin, 2017; Selden, 1997; Sowa & Selden, 2003). Therefore, the theory of active representation emphasises that a representative bureaucracy will advocate for the people it represents and be more responsive to their needs.

Bishu and Kennedy (2019), based on a content analysis of 92 journal articles on representative bureaucracy, concluded that the existing literature is mainly based on two dimensions of representativeness: gender and ethnicity. The authors also observed that the literature is geographically concentrated primarily in the context of the United States: 77% of the journal articles explore bureaucratic representativeness within this country.

Research Trends in Representative Bureaucracy

The literature on representative bureaucracy can be divided into three main waves: PR, passive to active representation and the link between passive and symbolic representation.

PR or Descriptive Representation (1970–1990)

This dimension was the primary focus of empirical studies on representative bureaucracy that emerged in the 1970s and continued until the 1990s. However, it has re-emerged as a ‘focus of study’—possibly due to SDG (as SDG focuses on inclusivity and responsive service delivery). Efforts are now being made to understand and explore how inclusive or representative the bureaucracy is.

This strand of literature mainly asks to what extent the bureaucracy mirrors the demographic composition of the country or the community it serves. In particular, the studies in this area examine whether the bureaucracy reflects the racial or gender composition of the country (national level) or the districts (sub-national or local level). In some cases, the officials’ socio-economic status has been considered to see whether they adequately represent the economic classes they serve. In analysing the socio-economic status of the officials, studies have mainly used the following features—occupation of parents (of the officials), income of the parents and educational level of the parents.

Passive to Active Representation (1990–Present)

Several studies show that PR can translate into active representation under certain conditions. Studies explore whether PR leads to active representation in the educational sector, health services, law enforcement and emergency service provisions. It is important to note that in the case of representative bureaucracy (and studies related to passive to active representation), the goal is to explore whether ‘greater representativeness’ results in ‘greater access’ for the underserved groups. In particular, the studies ask whether a change in representativeness results in more responsive service delivery or more allocation for underserved groups, given that public officials are supposed to support clients or constituents like them. In addition, the studies also try to discover the factors that allow PR to materialise into active representation.

Representative Bureaucracy in Bangladesh

In Bangladesh, very few studies have been conducted to explore the degree or nature of passive representativeness, and no study has been undertaken so far to explore whether passive representativeness may indeed result in active representation. Since 2000, several studies attempted to measure the degree of representativeness of women in the civil service of Bangladesh. The earlier studies in this domain concentrated on the ‘glass ceiling concept’ and explained in detail the challenges women face to be a part of the civil service. These studies mostly drew attention to the low share of female officials within the civil service. They explored how limitations within the public personnel management system of Bangladesh led to ‘…discrimination in recruitment, placement, advancement, mobility and training’ (Zafarullah, 2002).

Later studies, however, focused more on female officials’ representativeness and acknowledged that women’s representation in higher civil service was gradually increasing. However, these studies did not provide a systematic analysis and could not identify the female officials’ origin (i.e., their educational or socio-economic background (Sultan & Jahan 2016; Ahmed & Jahan 2018).).

Methodology and Data

The study explores socio-economic diversity in bureaucracy, focusing exclusively on the officials belonging to the administration cadre. This study’s main research questions are examined using primary and secondary data from various sources. For quantitative analysis, the study uses two different data sources—the database of training recipients at the BCS Administration Academy (BCSAA) and primary data from a survey of officials.

The study mainly focuses on administration cadre officials who were recruited after 2000. Between 2001 and 2020, recruitment notifications were published for 20 BCS examinations, 15 of which corresponded to administration cadre officials.

The BCSAA dataset contains information on 3,164 admin officials who were recruited from the 20th to 35th BCS examinations (excluding the 23rd and 32nd) and received training at BCSAA. This BCSAA database contains basic information about training participants, including their characteristics, educational qualifications and parents’ education and occupation. However, one limitation of the BCSAA dataset is that it does not contain information about the socio-economic status of the bureaucrats during their primary and secondary schooling stage, whether they lived in rural or urban areas, or whether they had access to electricity or TV during their childhood.

A survey was conducted for a sample of 209 officials to collect additional information about the administration cadre officials. The survey was implemented in two stages. In the first stage, primary data were collected from a new cohort of BCSAA training participants to include additional batches. The survey of training participants consists of 100 officials, 95 of whom belong to 35th–37th batches including 76 officials from the 37th BCS. In the second stage, the survey was conducted on a sample of officials from the 20th to 34th BCS batches, which provided information and responses from 109 officials. The sample was distributed across different batches of BCS officials to ensure representation from different batches. The sampling frame includes the list of government officials currently working as deputy, senior assistant and assistant secretaries in various ministries. Descriptive and analytic statistics are constructed using primary and secondary data to evaluate the trend and evolution of representativeness in the bureaucracy.

Results and Findings

Gender Diversity in BCS Administration

This section shows the status and trend in gender diversity, an indicator of demographic diversity, based on the analysis of recruitment notifications of MoPA. Figure 1 shows the percentage of female officials among administration cadre officials from 2001 to 2021. Within these 20 years, in the first five recruitment notifications (21st–27th), the average percentage of females was 24%, which increased to 31% in the second five (28th–33rd) and reached 33% in the last five recruitments. The percentage of females reached its lowest at 21% in the 22nd BCS, the leading recruitment notification published in November 2003. From the 22nd to the 30th BCS, female representativeness increased consistently to 35%. Representation of females reached its highest at 38% in the 34th BCS, the results of which were notified in May 2016. However, it is worth noting that female representation has dropped to 28% in the 37th BCS, which increased to 31% in the 38% BCS. This phenomenon may be due to increased competition in the BCS examination after the initiation of the pay scale in 2015. The findings, nevertheless, indicate that from a gender perspective, the administration cadre of the civil service of Bangladesh has become more representative over the years.

Figure 1.

Source: Authors’ calculation based on recruitment notifications of Ministry of Public Administration (MoPA).

As discussed earlier, if bureaucracy reflects the diversity of the society in terms of attributes of its population (gender, race, religion, etc.), then it can be identified as the presence of PR in bureaucracy. In the above discussion, it is observed that the representation of women has increased on average in the last decade compared to the preceding one. However, the proportion of women is still much below its proportion in the general population.

Table 1 presents the representativeness of females based on BCSAA and survey data. The results for the 37th BCS batch are based on the survey data of officials interviewed during a BCSAA training programme. From the 20th to the 36th BCS, the results show the representation based on BCSAA data that covers the training participants in earlier batches. In the case of the 36th BCS, information was available for only 53 officials. Table 1 also summarises the primary dataset of this study, which has been used to understand the long-term trend in representation.

Table 1.

Distribution of Administrative Officials by Gender (BCSAA & Survey Data).

BCS Batch	Female	%	Male	%	Total (N)	%
20	53	18.7	230	81.3	283	100
21	35	19.8	142	80.2	177	100
22	59	21.2	219	78.8	278	100
24	80	24.7	244	75.3	324	100
25	48	24.5	148	75.5	196	100
27	73	28.2	186	71.8	259	100
28	51	26.3	143	73.7	194	100
29	68	28.7	169	71.3	237	100
30	91	34.5	173	65.5	264	100
31	68	30.1	158	69.9	226	100
33	86	35.3	158	64.8	244	100
34	97	38.8	153	61.2	250	100
35	74	32.0	157	68.0	231	100
36	11	20.7	42	79.3	53	100
37*	21	27.6	55	72.4	76	100
Total	904	27.89	2,337	72.11	3,241	100

Note: *Data for the 37th batch are collected from a survey of training participants at the BCSAA.

Source: Authors’ calculation based on BCSAA & Survey data

Table 2 shows the representation by period, with officials from five different groups of BCS batches in each period. From Table 2, it is clear that the representation of female officials increased by around 5% on average from the second to fourth period. But the share of females declined by around 5% in the fifth period (35th–37th). This result raises the question of whether the pay scale hike is the main reason for this downturn in female representation from the 35th Batch.

Table 2.

Distribution of Officials by Gender and Period (BCSAA & Survey Data).

BCS Batches	Female (N)	%	Male (N)	%	Total (N)
20–22	147	19.9	591	80.1	738
24–25, 27	201	25.8	578	74.2	779
28–30	210	30.2	485	69.8	695
31, 33–34	251	34.9	469	65.1	720
35–37	108	30.0	252	70.0	360
Total	917	27.9	2,375	72.1	3,292

Source: Authors’ calculation based on BCSAA & Survey data.

Informational or Educational Diversity in BCS Administration

This section shows the pattern of representativeness in terms of the educational background of administration cadre officials. Using the combined dataset, Table 3 presents the overall status and pattern of changes in the distribution of admin officials in terms of their area of specialisation in higher studies. It appears that a significant proportion of the BCS officials completed their university-level education in subjects that belong to the Science discipline (42%), of which 17.1% studied Biological Science, 18.3% Science (others) and 6.6% Engineering or Computer Science (ECs). The rest of the officials completed their university-level education in Social Sciences (23.8%), Arts (23.1%) and Business Studies (11.3%).

Table 3.

Educational Background (Main Discipline) of BCS Administration Officials.

	Group of BCS Batches
	(1)	(2)	(3)	(4)	(5)
HE Subject	20–22	24–25,27	28–30	31,33–34	35–37	Total
Arts	21.3	25.2	28.5	21.9	13.8	23.1
Biological Science	15.7	16.5	18.1	18.5	16.6	17.1
Business Studies	9.4	12.9	11.2	10.6	13.2	11.3
Engineering/CS	3.0	5.2	5.0	7.9	17.2	6.6
Science (Others)	25.1	15.6	13.5	17.4	20.9	18.3
Social Science	25.6	24.7	23.6	23.7	18.3	23.8
Total	100.0	100.0	100.0	100.0	100.0	100.0

Source: Authors’ calculation based on BCSAA & Survey data.

Regarding the proportion of representation in each group of BCS batches, an exciting scenario emerges in the diversity of representation. In particular, a significant shift is observed in the distribution after the fourth period, from the 35th batches onward. The share of officials from Arts has increased steadily in the first three groups of BCS batches from the 20th to 30th BCS. However, a sharp decline in the share of officials with Arts background has been observed in the 4th and 5th group of officials. Although the share of officials from the Social Science discipline fluctuated around 24% between the 20th and 34th BCS (periods 1–4), it has dropped to 18.3% in the 5th group containing the officials of the 35th to 37th batches. A reverse trend is visible in the case of the share of officials having a science background, which increased from 43.9% in the fourth group to 54.7% in the 5th group. This jump is mainly driven by the increase in representation of officials having Engineering or Computer Science (ECs) background, where the share of this group more than doubled (7.9%–17.2%) from the 4th to 5th period. The above findings suggest a significant shift in the distribution of BCS admin officials has occurred after the introduction of the pay scale in 2015. The analysis below also found that the shift in academic discipline of recruited officials after the introduction of the pay scale occurred with a simultaneous change in the type of university.

Table 4 shows the distribution of recruited BCS officials according to the type of university from which they completed their degrees before joining BCS¹. Given the available data, a significant change is observed in the representation of officials from Dhaka University and that of Engineering University (which includes all the engineering universities of the country). First, the share of Dhaka University declined from 60.8% in the first group (20th–22nd) of BCS to 48.5% in the 2nd group (24th, 25th, 27th) of BCS and fluctuated around 50% from the 24th to 34th BCS. Second, the representation of Dhaka University in BCS recruitment has dropped significantly from 51.1% in the 4th group to 40.3% in the 5th group.

Table 4.

Educational Background (Type of University) of BCS Administration Officials.

	Group of BCS Batches
	(1)	(2)	(3)	(4)	(5)
HE University	20–22	24–25,27	28–30	31,34	35–37	Total
Agricultural University	3.9	5.7	9.2	5.5	7.2	6.2
Chittagong University	5.1	6.9	6.0	5.5	7.2	6.1
Dhaka University	60.8	48.5	49.1	51.1	40.3	51.0
Engineering University	2.7	4.2	3.0	3.8	12.8	4.5
Jahangirnagar University	5.0	6.0	5.0	7.6	5.6	5.7
Medical College	0.1	0.1	1.0	0.8	2.8	0.8
National University	10.0	15.3	9.5	8.8	5.3	10.5
Not available	0.4	0.5	0.6	0.4	0.8	0.5
Others	0.8	0.5	0.0	0.2	0.0	0.4
Private University	0.7	0.9	2.2	1.1	1.9	1.3
Public University (others)	5.4	6.8	6.6	8.8	13.6	7.5
Rajshahi University	5.0	4.5	7.8	6.5	2.5	5.4
Total	100.0	100.0	100.0	100.0	100.0	100.0

Source: Authors’ calculation based on BCSAA & Survey data.

On the other hand, the representation of the Engineering University category depicted a sharp increase from just 3.8% in the 4th group to 12.8% in the 5th group of officials. Interestingly, although the representation of Dhaka University has declined after the introduction of the pay scale in 2015, the proportion of students from other public universities has increased from 8.8% to 13.6% over the same period. The above findings suggest that the introduction of the 8th pay scale is likely to contribute to improving the informational diversity of the administration cadre. In the next section, using a regression framework, we formally test whether the introduction of the 8th pay scale has increased the informational diversity of the administration cadre.

The 8th Pay Scale and Informational Diversity

In Tables 3 and 4, we observe that the representation of officials from Engineering and Computer Science backgrounds was the lowest before the introduction of the pay scale. Thus, we start our analysis with specification (1), where we test the impact of the 8th pay scale on the representation of this group compared to all other disciplines. The primary regression specification is as follows:

l n E m p_{i t} = α + τ_{t} + δ_{1} E C s + β_{1} E C s \times P S_{t} + ϵ_{i t}

(1)

where $l n E m p_{i t}$ is the log of the number of officials with a higher education in subject $i$ (or from university $i$ ), in BCS batch $t$ . For example, $E m p_{E C s 20}$ indicates the number of officials in the 20th BCS batch who completed their graduation in ECs discipline (or from an engineering university). $E C s$ is a dummy variable representing Engineering or Computer science discipline (or engineering university). $P S_{t}$ indicates the year after the introduction of the 8th pay scale, taking the value 1 from the 35th to the 37th BCS. $τ_{t}$ is a set of BCS batches or time dummies.

To explore whether the introduction of the 8th pay scale has also affected the representation of officials of any other discipline, we modify specification (1) by including a set of discipline-specific intercepts $α_{i}$ and the interaction between each discipline specific dummy and the post pay scale dummy PS_t.

l n E m p_{i t} = α_{i} + τ_{t} + β_{1} E C s \times P S_{t} + β_{2} A R \times P S_{t} + β_{3} B S \times P S_{t} + β_{4} B i o S \times P S_{t} + β_{5} O S c \times P S_{t} + ϵ_{i t}

(2)

where AR, BS, BioS and OSc denote Arts, Business Studies, Biological Science and Other Sciences, respectively, and Social Sciences is used as a base category. Table 5 shows the regression results for subject level analysis. In column (1), the coefficient on Engineering or CS (−1.301) and that on the interaction term (1.395) indicate that the introduction of the 8th pay scale has a strong positive impact on the recruitment of admin officials from the ECs discipline.

Table 5.

Effects of Pay Scale on the Recruitment of Officials by Academic Discipline.

	(1)	(2)	(3)
Engineering or CS	−1.301***	−1.604***	−1.604***
	(0.124)	(0.128)	(0.129)
Arts		0.00	0.00
		(0.085)	(0.086)
Business Studies		−0.826***	−0.826***
		(0.089)	(0.090)
Biological Science		−0.365***	−0.365***
		(0.074)	(0.075)
Science Others		−0.327***	−0.327***
		(0.109)	(0.110)
Engineering or CS × PS	1.395***	1.549***	1.457***
	(0.240)	(0.334)	(0.129)
Arts × PS		−0.487	−0.147*
		(0.319)	(0.086)
Business Studies × PS		0.356	0.597***
		(0.304)	(0.090)
Biological Science × PS		0.217	0.318***
		(0.272)	(0.075)
Science Others × PS		0.681	0.04
		(0.412)	(0.110)
BCS Batch (year) Dummies	Yes	Yes	Yes

Constant	3.879***	4.182***	4.182***
	(0.141)	(0.134)	(0.136)
R-squared	0.701	0.85	0.801
N	90	90	78

Note: The log of the number of officials with a degree in the subject iis the dependent variable. All regression used BCS batch-specific dummy variables. Robust standard errors are reported in parentheses. *p < .10, **p < .05, ***p < .01.

In column (2) of Table 5, using Social Sciences as the base category, we observe that the admin cadre has become more informationally diverse with the introduction of the 8th pay scale by raising the number of officials with higher education in ECs, BS and BioS. However, the results of the last two groups are not statistically significant. As the number of observations used for calculating the number of officials by discipline is significantly small for the 36th and 37th batches, we exclude these two batches in column (3). Now, the interaction coefficients of BS and BioS are statistically significant.

Table 6 presents the results of the representation of administrative officials by university. In line with the findings in Table 5, there was a significant jump in the number of officials from engineering universities in the batches recruited after the pay scale’s introduction. Interestingly, the representation of other public and private universities has also increased after the introduction of the pay scale compared to Dhaka University, while the representation of National University has declined relative to the base category.

Table 6.

Effects of 8th Pay Scale on the Recruitment of Officials in terms of University.

	(1)	(2)	(3)
Engineering	−0.391**	−2.584***	−2.584***
	(0.151)	(0.138)	(0.140)
Agricultural		−2.109***	−2.109***
		(0.134)	(0.136)
Chittagong		−2.135***	−2.135***
		(0.132)	(0.134)
Jahangirnagar		−2.159***	−2.159***
		(0.138)	(0.139)
Rajshahi		−2.120***	−2.120***
		(0.131)	(0.132)
Public Others		−1.980***	−1.980***
		(0.125)	(0.126)
National		−1.546***	−1.546***
		(0.136)	(0.138)
Private		−3.590***	−3.590***
		(0.198)	(0.200)
Medical		−4.102***	−4.102***
		(0.199)	(0.201)
Engineering × PS	1.074**	1.572***	0.984***
	(0.412)	(0.411)	(0.140)
Agricultural × PS		0.484***	0.237*
		(0.184)	(0.136)
Chittagong × PS		0.37	0.535***
		(0.453)	(0.134)
Jahangirnagar × PS		0.429**	0.153
		(0.186)	(0.139)
Rajshahi × PS		−0.525*	−0.327**
		(0.289)	(0.132)
Public Others × PS		0.994***	0.770***
		(0.229)	(0.126)
National × PS		−0.219	−0.614***
		(0.363)	(0.138)
Private × PS		1.116***	0.555***
		(0.278)	(0.200)
Medical × PS		1.836***	1.404***
		(0.272)	(0.201)
BCS Batch (year) Dummies	Yes	Yes	Yes
Constant	2.508***	4.702***	4.702***
	(0.439)	(0.166)	(0.168)
R-squared	0.036	0.877	0.88
N	150	150	130

Note: The log of the number of officials having a degree from a university is used as the dependent variable. All regression used BCS batch-specific dummy variables. Robust standard errors are reported in parentheses. *p < .10, **p <.05, ***p < .01.

Diversity in Socio-economic Background

This section analyses the socio-economic background of the BCS officials, focusing on the fathers’ educational qualifications and occupations².

Table 7 shows the education level of the fathers of BCS officials³. Interestingly, the representation of officials whose fathers’ education levels were below undergraduate (Below Secondary, SSC and HSC) has increased from 43.5% in the 4th group to 49.0% in the 5th group. In contrast, the representation of officials having more than higher secondary level education has decreased slightly from 48.7% to 47.9%. Although the share of missing observations has dropped simultaneously by 4.71% between the two periods, Table 7 still indicates a move towards an increase in representation from less well-off socio-economic backgrounds.

Table 7.

Education Level of the Officials’ Fathers.

	Group of BCS Batches
The education level of fathers	(1)	(2)	(3)	(4)	(5)
The education level of fathers	20–22	24–25,27	28–30	31,34	35–37	Total
Below Secondary	15.2	10.0	8.1	10.9	14.2	11.5
SSC	18.1	13.0	13.1	14.7	15.9	14.9
HSC	12.6	17.1	18.9	17.9	18.9	16.7
BA/BCom/BSc	29.2	34.3	35.1	30.7	30.4	32.2
Masters or above	17.9	16.3	16.0	18.1	17.6	17.0
NA	7.0	9.4	8.9	7.8	3.1	7.7
Total	100	100	100	100	100	100

Table 8 shows the regression results using the log of the number of officials by educational qualifications of fathers as the dependent variable. In column (1) of Table 8, using the full sample (excluding the 33rd batch), we observe that the coefficient on the interaction term between the Below Secondary and $P S_{t}$ is positive and statistically significant at 10% level, whereas the interaction coefficient corresponding to SSC is positive but statistically insignificant. As the representations of these two groups were the highest in the first period (20th–22nd BCS), we show the results by excluding these three batches in column (2), where both the interaction coefficients appear to be positive and statistically significant. Therefore, based on the regression results, the 8th pay scale positively impacts the representation of officials from less well-off socio-economic backgrounds.

Table 8.

Effects of Pay Scale on Socio-economic Diversity (by Fathers’ Education).

	(1)	(2)
Below Secondary	–0.437*** (0.126)	–0.589*** (0.129)
SSC	–0.123 (0.102)	–0.237** (0.110)
Below Secondary × PS	0.588* (0.335)	0.740** (0.339)
SSC × PS	0.458 (0.335)	0.572* (0.340)
Constant	3.904*** (0.199)	4.025***
R-squared	0.392	0.427
N	84	66

Note: The dependent variable shows the log of the number of officials whose fathers’ educational qualifications fall under level i. All regression used BCS batch-specific dummy variables. Robust standard errors are reported in parentheses. *p < .10, **p < .05, ***p < .01.

Table 9 shows the occupational status of the officials’ fathers, which is a key indicator of the socio-economic status of the civil servants. A related question of interest is whether the representation of officials from rural backgrounds has changed with the introduction of the pay scale. Although the BCSAA dataset does not contain information regarding the locational background of the officials, we can get some indication regarding the change in representation from rural backgrounds by analysing the occupational background of the officials’ fathers. In particular, we observe that the proportion of officials whose fathers were farmers declined steadily in the first four periods, from 20th to 34th BCS, but increased to 14.8% in the 5th period from 12.0% in the 4th period. This is consistent with the above findings that the representations of officials whose fathers’ education was less than undergraduate level have increased over the same period. The above results provide some evidence supporting the argument that the increase in pay scale has encouraged individuals from less well-off backgrounds or rural locations to join the civil service. While the share of officials whose fathers were involved in government services has shown an increasing trend from the first to fourth period, it has declined after the introduction of the pay scale.

Table 9.

Occupation of the Officials’ Fathers.

	Group of BCS Batches
	(1)	(2)	(3)	(4)	(5)
Fathers’ Occupation	20–22	24–25,27	28–30	31,34	35–36	Total
Banker	1.2	2.4	2.7	2.5	3.2	2.3
Business	17.9	15.4	17.0	22.7	29.6	18.9
Doctor	1.2	0.6	1.0	1.1	1.4	1.0
Engineer	0.4	0.1	1.3	0.0	0.0	0.4
Farmer	19.9	16.9	14.4	12.0	14.8	16.1
Government Service	33.6	34.7	39.4	40.6	36.6	36.6
Lawyer	2.4	4.6	2.0	2.3	1.4	2.8
Not available	4.3	5.8	5.3	6.5	2.5	5.1
Others	0.7	0.4	0.0	0.0	0.4	0.3
Private Service	1.8	3.7	0.9	0.0	0.7	1.7
Teacher	16.6	15.3	16.0	12.4	9.5	14.8
Total	100.0	100.0	100.0	100.0	100.0	100.0

Source: Authors’ calculation based on BCSAA & Survey data.

As the business category includes various activities, it is not clear whether an increase in representation from this category indicates an increase in the share of more well-off or less well-off households. By separating the business category occupation in terms of educational attainment of the fathers, it is observed that the proportion of officials whose fathers were involved in business and achieved SSC or below level of education increased in both the 4th and the 5th period, while the business category with HSC or above level of education has increased mainly in the 5th period.

The effects of the 8th pay scale on the distribution of officials by occupational status of their fathers are presented in Table 10. Column (1) uses data from the 20th to 36th BCS and column (2) uses data from the 24th to 36th BCS batches, excluding the 33rd batch. In both columns, government service is considered as the base category. It is observed that the representation of most of the key socio-economic classes (occupational status of fathers) has increased after the introduction of the pay scale, except for teachers and other categories. However, only the interaction term between Business (SSC or Below) and the pay scale dummy is statistically significant. Overall, the results provide evidence in favour of increased representation from lower socio-economic classes.

Table 10.

Effects of Pay Scale on Socio-economic Diversity (Fathers’ Occupation).

	(1)	(2)
Business (SSC or Below)	−1.614***	−1.762***
	(0.165)	(0.205)
Business (HSC or Above)	−1.292***	−1.237***
	(0.148)	(0.189)
Doctor or Engineer	−3.062***	−3.138***
	(0.188)	(0.211)
Farmer	−0.845***	−0.937***
	(0.103)	(0.103)
Private Service	−3.496***	−3.480***
	(0.338)	(0.392)
Teacher	−0.885***	−0.944***
	(0.121)	(0.102)
Others	−1.273***	−1.196***
	(0.122)	(0.116)
Business (SSC or Below) × PS	0.736**	0.884***
	(0.295)	(0.324)
Business (HSC or Above) × PS	0.386	0.331
	(0.538)	(0.564)
Doctor or Engineer × PS	0.477	0.553
	(0.803)	(0.830)
Farmer × PS	0.152	0.244
	(0.266)	(0.272)
Others × PS	−0.04	−0.117
	(0.275)	(0.278)
Private Service × PS	0.421	0.404
	(0.417)	(0.466)
Teacher × PS	−0.277	−0.218
	(0.289)	(0.289)
Constant	4.819***	3.070***
	(0.240)	(0.277)
R-squared	0.819	0.83
N	104	80

Note: The dependent variable is the log of the number of officials whose fathers’ occupation falls in group $i$ . All regression used BCS batch-specific dummy variables. Robust standard errors are reported in parentheses. *p < .10, **p < .05, ***p < .01.

Diversity in Socio-economic Background: Findings from the Survey Data

In the previous section, the trend in representativeness was observed for a range of respondents’ characteristics, including the officials’ educational backgrounds and the education and occupation of their fathers. However, a range of other socio-economic backgrounds remains unknown. This section focuses on a range of socio-economic characteristics of the respondents along with their educational background based on the survey of civil service officials. The sample is divided into two parts before and after the introduction of the pay scale in 2015: officials of 20th–34th BCS Batches and 35th–37th BCS batches.

Characteristics of the Survey Respondents

Out of a sample of 209 officials, 54.5% belong to the 20th–34th BCS group and 45.5% belong to the 35th–37th BCS group (Table 11). In terms of gender, around 34.9% of the respondents are female and 65.1% are male (Table 11).

Table 11.

Distribution of Respondents by Gender and BCS Batches.

BCS_Gr_SVY1	Category	Freq.	Percent	Category	Frequency	Percent	Total
20th–34th BCS	Overall	114	54.5	BCS 35–37	95	45.5	209
Full Sample	Female	73	34.9	Male	136	65.1	209
20th–34th BCS	Female	43	37.7	Male	71	62.3	114
35th–37th BCS	Female	30	31.6	Male	65	68.4	95

Source: Authors’ calculation based on Survey data.

Based on the combined dataset, the share of females in the 35th–37th group is 30%, which is slightly lower than the estimated share of 31.6% from survey data. Conversely, the share of females in the survey data for the 20th–34th BCS group was 37.7%, which is significantly higher than the corresponding figure from the combined dataset. Therefore, the proportion of the females in the survey data is higher than the actual representation of females in the BCS administration.

The sample is quite balanced in terms of the respondents’ ages. In Table 12, the average age of the females is 37.3, and that of the males is 38.0.

Table 12.

Average Age of the Respondents.

BCS_Gr_SVY1	Category	Average Age	Category	Average Age
Full Sample	Female	37.3	Male	38.0
BCS 20–34	Female	41.2	Male	43.6
BCS 35–37	Female	31.7	Male	31.8
BCS 20–34	Overall	42.7	BCS 35–37	31.8

Source: Authors’ calculation based on Survey data.

The average age of the female respondents (41.2) is slightly lower than that of males (43.6) in the 20th–34th BCS group, whereas for the 35th–37th BCS group, the average age of the two groups is almost equal (31.7 and 31.8 years). In terms of age group, 93.7% of the respondents in the 35th–37th BCS group fall within the range of 29–34 years, whereas in the 20th–34th BCS group, the proportion of respondents is highest in the 41–46 years group (Figure 2).

Figure 2.

Source: Authors’ calculation based on Survey data.

Representation from Rural Areas

The respondents were asked whether they lived in an urban or rural area during their primary and secondary education, separately. In the case of primary education, the representation of officials from urban and rural areas has changed only marginally between the two BCS groups.

Figure 3 shows that in the 20th–34th BCS group, 47.2% of the respondents lived in rural areas during their primary education level, which is 48.4% for the 35th–37th BCS group (Figure 4). However, in the case of secondary education, the share of respondents who lived in rural areas is marginally lower for the 35th–37th BCS group compared to the 20th–34th BCS group. Therefore, the location-wise direction of change in the distribution after the introduction of the pay scale is not conclusive from the survey data. Thus, the introduction of the pay scale does not affect the representation of officials with rural backgrounds. This is consistent with our earlier analysis; we observed that the introduction of a pay scale has no statistically significant impact on the representation of officials whose fathers were farmers.

Figure 3.

Source: Authors’ calculation based on Survey data.

Figure 4.

Source: Authors’ calculation based on Survey data.

Access to Television During Childhood

Access to Television

Two interesting facts emerge from Figures 5 and 6, which show the percentages of officials who had access to TV during their primary and secondary education, respectively. First, access to TV has increased from primary to secondary education level for both the 20th–34th and 35th–37th BCS groups. This feature is most likely to result from increased access to TV at the country level over time. Second, the proportion of officials who had access to TV in primary or secondary education has increased in the 35th–37th group compared to the 20th–34th BCS group. However, based on this finding, it cannot be concluded that the representative of more well-off households has increased because access to TV has increased in the nation as a whole over time, reflecting the upgradation of the wealth status of the general households.

Figure 5.

Source: Authors’ calculation based on Survey data.

Figure 6.

Source: Authors’ calculation based on Survey data.

Educational Background of the Respondents

In our earlier analysis based on the combined dataset, we observed that the proportion of officials with Engineering or Computer Science backgrounds increased significantly (more than doubled) after the introduction of the pay scale in 2015. Table 13, based on the survey data only, echoes the same conclusion, where the share of officials with ECs background is around 20% for the 35th–37th group, which was just 1.8% for the 20th–34th BCS group. On the other hand, the share of officials with Arts background has decreased significantly after the introduction of the Pay Scale.

Table 13.

Main Subject of Study in Higher Education (in Percent).

HE	BCS 20–34	BCS 35–37	Total
Arts	32.43	12.6	23.3
Biological Science	0	11.6	5.34
Business Studies	15.32	13.7	14.56
Engineering/CS	1.8	20.0	10.19
Science (others)	38.74	32.63	35.92
Social Science	11.71	9.47	10.68
Total	100	100	100

Source: Authors’ calculation based on Survey data.

The proportion of officials whose fathers had Masters or higher education level is around 21% for both the 20th–34th and 35th–37th BCS groups (Figure 7). However, an interesting shift has been observed in other education levels. The proportion of officials with fathers having BA/BCom/BSc has declined from 46.0% in the 20th–34th to 37.9% in the 35th–37th BCS group, while the share of SSC/HSC group has increased from 19.5% to 30.5% in the second group. The above findings suggest that the representativeness of officials with less well-off socio-economic status has increased in the BCS administration cadre.

Figure 7.

Source: Authors’ calculation based on Survey data.

Conclusion

The importance of diversity in bureaucracy is well-recognised in the global literature on public administration and public policy. At the same time, there is a wider agreement that as the bureaucracy becomes more representative, the entity becomes more sympathetic to the needs and concerns of the marginalised groups, religious, ethnic and racial minorities, and tries to be more responsive to address citizens’ demands. Unfortunately, however, in the context of Bangladesh, the utility of representative bureaucracy has remained largely ignored, and we have a minimal idea of how representative our bureaucracy is or whether the current level of representativeness can effectively translate into active representation. From this perspective, this study plays an important role in filling the gap in our current state of knowledge.

This study focuses on three key dimensions of diversity in bureaucracy: demographic, informational and socio-economic. By analysing the data on officials of the administration cadre services over 20 years, in this study, we explore how the composition of the civil service is gradually changing and how representative it has become over time. Our study shows that contrary to the perceptions championed by the earlier studies and popular perception, the administration cadre service experienced significant changes during the study period, and the 8th pay scale played an important role in increasing the diversity in civil service. In the case of gender (demographic) diversity, we found that the representation of female officials has gradually increased over time but showed a slight downward trend in the later periods. The bureaucracy has become more diverse in terms of the educational background of the officials. We found that the share of civil servants with an Engineering and Computer Science background almost doubled during the 35th–37th BCS compared to the 31st–34th BCS. In contrast, the share of officials with Arts and Social Science backgrounds has dropped significantly over the same period. Based on the findings, it can be argued that the distribution of BCS admin officials has changed after the introduction of the pay scale in 2015.

The main indicators of socio-economic background also show noticeable changes over the study period. The representations of bureaucrats with fathers’ having below undergraduate (Below Secondary, SSC and HSC) level education has increased significantly in the last period. The distribution of officials based on fathers’ occupation has shown interesting changes over time. One noticeable observation is that the share of officials whose fathers were farmers declined gradually in the first four periods but increased in the fifth period, that is, after the introduction of the pay scale, while the scenario is just the opposite for the officials whose fathers were involved in government services. However, based on the regression analysis, the 8th pay scale has no statistically significant impact on the representation of officials whose fathers were farmers. In the survey of officials, we also found that the representation of officials who lived in rural areas during their primary or secondary education has not been affected by the introduction of the pay scale. Overall, we observe that the introduction of the pay scale positively affects the diversity in bureaucracy by increasing the representation of officials from less well-off socio-economic backgrounds.

However, it must be acknowledged that the scope of the current study has been limited. We have focused only on the administration cadre services. Still, a more extensive study encompassing the entire bureaucracy is required to understand the actual level of PR and the possibility of translating PR into active one. At the same time, in this study, we have considered gender, educational background and socio-economic background (measured through the civil servants’ fathers’ occupation and education) to measure the level of PR. Still, due to a smaller sample size, we could not consider religious or ethnic representation. Further study is required to capture all these concepts. Furthermore, a large-scale survey-based study is needed to understand the change in value structure within the civil service, which will evaluate the changes taking place within the civil service starting from 1972 and not just in the last 20 years. These are some key areas where further research is necessary.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors are grateful to the Bangladesh Civil Service Administration Academy for funding the study titled “Representative Bureaucracy and Development Administration.”

Notes

References

Ahmed

N. A.

, & Jahan

(2018). Balancing work and family: Women in Bangladesh civil service. In

Ahmed

(ed.) Women in governing institutions in South Asia (pp. 141–162). Routledge.

Baniamin

H. M.

, Jamil

, & Askvik

(2020). A mismatch between lower performance and higher trust in the civil service: Can culture provide an explanation? International Political Science Review , 41(2), 192–206.

Bowling

C. J.

, Kelleher

C. A.

, Jones

, & Wright

D. S.

(2006). Cracked ceilings, firmer floors, and weakening walls: Trends and patterns in gender representation among executives leading American state agencies, 1970–2000. Public Administration Review , 66(6), 823–836.

Choi

, Hong

, & Lee

J. W.

(2018). Does increasing gender representativeness and diversity improve organizational integrity? Public Personnel Management , 47(1), 73–92.

Huque

A. S.

(1997). The impact of colonialism: Thoughts on politics and governance in Bangladesh. Asian Affairs , 28(1), 15–27.

Huque

A. S.

, & Rahman

M. T.

(2003). From domination to alliance: Shifting strategies and accumulation of power by the bureaucracy in Bangladesh. Public Organization Review , 3(4), 403–418.

Jahan

(2006). Public administration in Bangladesh (Working Paper Series, 1). Centre for Governance Studies.

Keiser

L. R.

, Wilkins

V. M.

, Meier

K. J.

, & Holland

C. A.

(2002). Lipstick and logarithms: Gender, institutional context, and representative bureaucracy. American Political Science Review , 96(3), 553–564.

Kennedy

(2014). Unraveling representative bureaucracy: A systematic analysis of the literature. Administration & Society , 46(4), 395–421.

10.

Kingsley

J. D.

(1944). Representative bureaucracy: Classic readings and continuing controversies (pp. 12–18). Routledge.

11.

LeRoux

(2009). The effects of descriptive representation on nonprofits’ civic intermediary roles: A test of the ‘racial mismatch’ hypothesis in the social services sector. Nonprofit and Voluntary Sector Quarterly , 38(5), 741–760.

12.

Levitan

(1946). The responsibility of administrative officials in a democratic society. Political Science Quarterly , 61, 562–598.

13.

Long

N. E.

(1952). Bureaucracy and constitutionalism. American Political Science Review , 46, 808–818.

14.

Moldovan

(2016). Representative bureaucracy in Romania? Gender and leadership in central public administration. Gender and Leadership in Central Public Administration (October 18, 2016). Transylvanian Review of Administrative Sciences , (48E), 66–83.

15.

Mosher

F. C.

(1968). Democracy and the public service (vol. 53). Oxford University Press.

16.

Parthasarathy

, Doboli

, & Paulus

P. B.

(2011). Entrepreneurship. In Runco

and Pritzker

(Eds), Encyclopedia of creativity (2nd ed., vol. 1, pp. 461–467). Elsevier.

17.

Riccucci

N. M.

, & Van Ryzin

G. G.

(2017). Representative bureaucracy: A lever to enhance social equity, coproduction, and democracy. Public Administration Review , 77(1), 21–30.

18.

Selden

S. C.

(1997). Representative bureaucracy: Examining the linkage between passive and active representation in the farmers home administration. The American Review of Public Administration , 27(1), 22–42.

19.

Shahan

A. M.

, Jahan

, & Khair

(2021). A glimpse of light in darkness: Performance-based accountability in Bangladesh public administration. Public Administration and Development , 41(4), 191–202.

20.

Sowa

J. E.

, & Selden

S. C.

(2003). Administrative discretion and active representation: An expansion of the theory of representative bureaucracy. Public Administration Review , 63(6), 700–710.

21.

Sultan

& Jahan

(2016). Openings and resistances: Women in the Bangladesh civil service. In

Ahmed

(ed.), Public policy and governance in Bangladesh, forty years of experience (pp. 104–113). Routledge.

22.

Zafarullah

H. M.

(1987). Public administration in the first decade of Bangladesh: Some observations on developments and trends. Asian Survey , 27(4), 459–476.

23.

Zafarullah

, & Huque

A. S.

(2001). Public management for good governance: Reforms, regimes, and reality in Bangladesh. International Journal of Public Administration , 24(12), 1379–1403.