Policy capacity matters for capacity development: comparing teacher in-service training and career advancement in basic education systems of India and China

Capacity development is central to the study and practice of public policy and administration, but ensuring its effectiveness requires a substantial amount of policy capacity from government agencies tasked to design and implement it. Identifying the right mix of policy capacity that governments should possess has been made difficult due to conceptual and operational problems. This article addresses the gap by developing a framework that conceptualizes policy capacity as the ability of governments to perform analytical, operational, and political functions. Drawing on the results of an original teacher survey and complementary sources, the article shows that variations on different dimensions of policy capacity have led to significant differences in the effectiveness of capacity development initiatives, especially as perceived by teachers. Therefore, without understanding and catering to the needs of the targets whose capacity is supposedly being developed, capacity development initiatives meant to be supportive are likely to be dissatisfying and disappointing instead. Points for practitioners This article highlights the importance of policy capacity and further unpacks how its analytical, operational, and political dimensions are essential to the successful delivery of capacity development. Through a rich account of the comparative case of India and China, it illustrates that all these dimensions are important, without any one being a stand-alone panacea. Above all, it is important to pay attention to the recipients of capacity development programs; without doing so, top-down program delivery ignorant of their needs is likely to be poorly received despite the original intention of developing capacity.


Introduction 1
Capacity development (CD) is crucial in enabling individuals and institutions in developing countries to effectively design policies, enact programs, and deliver public services (Burgess, 1975;Hope, 2009). The experience of many developmental states in Asia and Latin America reinforced the importance of consciously constructing, nurturing, and maintaining the right mix of policy capacity to ensure that policy implementation produces its desired outcomes (Haque and Puppim de Oliveira, 2020). However, CD entails a complex process of coordination at various levels of the government, which often makes any effort to develop government capacity very challenging (El-Taliawi and Van Der Wal, 2019). Thus, delivering effective CD initiatives requires substantial policy capacity on the part of the government departments tasked to do so. However, understanding what sorts of policy capacities are actually needed by these government departments remains largely unexplored due to the conceptual and operational problems that Honadle (1981) earlier raised about studying capacity. A recent bibliometric analysis established the fragmentation of the capacity literature as a result of these problems and concluded that "capacity is widely researched but poorly theorized" (Saguin et al., 2018b: 10). To generate lessons for policymakers about CD, it is vital to further conceptualize and empirically examine the policy capacities needed to make CD initiatives effective.
This article seeks to address this gap by proposing the recent conceptualization of "policy capacity" as a means to advance the CD literature (Wu et al., 2015). Using CD for teachers as its empirical focus, the article proposes a framework built on a multi-level and nested understanding of policy capacity: that capacity of the policy system as a whole shapes the likelihood of achieving desirable outcomes for CD initiatives aimed at developing individual capacities. More specifically, the article examines how variation in governments' policy capacity can affect variation in CD outcomes, especially as viewed from the perspectives of individuals whose capacities are being developed. In the next section, we present the generalized framework that disaggregates policy capacity at the policy system level into its constitutive dimensions (analytical, operational, and political) and then use it to compare the CD initiatives for teachers in the two largest basic education systems in the world: India and China. Systematic comparisons suggest that variations on different dimensions of policy capacity in Beijing and Delhi have led to significant differences in the effectiveness of such arrangements as perceived by teachers, even though governments in both regions are almost equally engaged with teacher CD at first glance. Both regions' limited analytical capacity constrains the inclusion of teacher preferences for training but Beijing partially compensates for this capacity deficit by relying on a highly embedded network of academics and experts, including teacher experts recognized through the career advancement system. Operational capacity varies quite significantly, with Delhi's teacher training program suffering from inadequate coverage and unreliability, while Beijing's capacity is demonstrated through its comprehensive, targeted, and efficient provision of teacher training. Lastly, both regions have deployed political capacity to increase budgets for education, but unlike Beijing, Delhi is less able to benefit from legitimacy derived from including teachers' voices in CD planning.
This article contributes to the public policy and administration literature by demonstrating the usefulness of the policy capacity concept in analyzing CD programs in education. It also adds to the scant literature on policy capacity in developing countries (Saguin et al., 2018a) by scrutinizing policy capacity and CD experiences from two of the world's largest developing countries. In doing so, it offers valuable lessons to other developing countries facing similar CD challenges (e.g. Awortwi, 2010). Comparative studies of education policy in China and India have paid more attention to policy documents and outcomes from a top-down angle, which is understandable given their focus on the national or state/provincial levels (Dreze and Sen, 1995;Smith and Joshi, 2016;Yan, 2018). Our study supplements this focus by offering a much-needed bottom-up perspective of CD in the education sector, thanks largely to the use of an original teacher survey. Finally, the study enriches the discussions on teacher in-service training, which have presented a mixed record so far (see Evans and Popova, 2016;cf. Piper et al., 2018), by adding a more nuanced picture of how training as a CD measure is practiced and received through the theoretical lens of policy capacity.
Conceptualizing the policy capacity for CD in education CD forms part of the government's repertoire in improving the achievement of education outcomes (McDonnell and Elmore, 1987). However, many of the popular reform proposals over the last few decades, such as education decentralization and privatization, rarely recognize the need for CD in enabling those at the receiving end of greater autonomy and choice, including local governments, parents, and private schools, to perform additional educational functions. When individual capacity of these actors to absorb the governance roles delegated to them is not considered, even well-designed education policies tend to be ineffective and fail to be genuinely developmental (Saguin, 2019;Yan, 2019a).
CD effectiveness has been argued to be contingent on generating information about how education services are produced, how learning occurs, and how actors behave as they deliver educational services (Hannaway and Woodroffe, 2003;Saguin and Ramesh, 2020). Political capacity is also important because in driving "forward an effective and efficient development of basic education, a close collaboration between these more private actors and the more public actors is indispensable" (Hirosato and Kitamura, 2009: 48). Since governments must orchestrate the collaborative work with non-governmental actors, they must develop staff competencies and demonstrate strong political leadership (De Grauwe, 2009). While these works start to recognize the value of CD, the principles they suggest remain too contextual and too disparate to be generalizable, and do not show an integrated but nuanced understanding of what government departments responsible for designing and implementing CD interventions should do.
The policy capacity framework recently elaborated in the public policy field offers a promising direction to address this gap. Wu et al. (2015: 166) propose a generalized conceptualization of policy capacity, defined as "the set of skills and resources-or competences and capabilities-necessary to perform policy functions." Policy capacity is also comprised of analytical, operational, and political dimensions across the systemic, organizational, and individual levels (Ramesh et al., 2016). This view of policy capacity aligns with treating technical, operational, and political capabilities of public agencies as the critical elements of a successful development policy (Cornick et al., 2016). Umeh (1992) similarly emphasized the importance of these skills as they relate to planning, policy analysis, adaptation, and political maneuvering in development administration.
The application of this framework on understanding the levels of analytical, operational, and political capacities can help better unpack the relationship between policy capacity and effective CD as successful CD can be perceived as a product of an optimum combination of the dimensions of policy capacity. Insofar as CD in education is a crucial process in ensuring that stakeholders involved are capable of delivering satisfying education services, getting these different yet interrelated dimensions of policy capacity right is all the more important (see Figure 1 2 ).
Analytical capacity is broadly about generating intelligence about society. It requires a sophisticated "machinery and processes for collecting and analyzing data and organizational commitment to evidence-based policy" (Pattyn and Brans, 2015, cited from Wu et al., 2015: 169). In the context of CD, it refers to the ability to better specify the needs for and gaps in CD, which empowers governments to track the status and progress of CD. Whereas other aspects of education, such as infrastructure, finance, and, to some extent, student learning, can be easily gathered in more objective formats through education management information systems or data sets on student exam scores, analytical capacity for CD interventions is often reflected in a subjective manner through consultation with educational experts, school principals, and teachers.
Operational capacity can be understood as the ability to use appropriate policy instruments and arrangements to solve policy problems (Bali et al., 2019). While resource availability is an important factor to consider, operational capacity in CD necessitates that resources be effectively utilized through the design and implementation of operationally feasible instruments to cover the gaps identified. As "tools have targets" that are "expected to be affected by policy activity" (Howlett, 2018: 101), CD effectiveness on the operational dimension also requires minimization of exclusion bias in targeting. Governments with high operational capacity would further be able to deliver more relevant, tailored, and sustainable CD programs to better suit the differentiated and evolving needs within the overarching inclusiveness of coverage.
Political capacity refers to the public engagement resources and skills within the system that give it the competence to acquire and retain legitimacy, as well as the ability to reconcile political differences (Bali et al., 2019). Given that the education sector is a multi-stakeholder enterprise, aligning the diverse interests of stakeholders, eliciting their support, or at least mitigating political oppositions are even more important for achieving the ultimate goal of improving student learning (Mizala and Schneider, 2014;Schneider et al., 2019). As it relates to CD programs, political capacity can similarly be manifested in political will or commitment toward CD and the systematic pursuit of engaging different stakeholders. The latter can be accomplished by creating a platform for hearing the voices of different stakeholders and exchanging information and opinions, as well as by bringing them on board as partners in the design and delivery of CD reforms. For the stakeholders at the receiving end of CD measures, listening and catering to their needs, voices, and concerns through these engagement efforts would especially make them feel motivated and empowered to carry out their work.

Data and research methods
To explore how efforts in developing teacher capacity are put into practice and the policy capacity involved in delivering it, this study relies mainly on data collected through a teacher survey conducted in government middle schools in Beijing and Delhi between September 2016 and December 2017. As the capital cities of China and India, respectively, these two sites are purposively chosen as they are relatively more comparable in terms of CD efforts than their least advanced regions given the uneven education development in both countries. As such, the case selection suits the exploratory inquiries on this important yet under-examined topic, though we are aware that the two capital cities are far from being representative of India and China, for which future explorations may be extended to other regions facing varied CD demands and challenges.
An empirical focus is given to CD initiatives catered to teachers in order to rectify prior scholarly and policy focus at the organizational or systemic levels. How teachers deliver the curriculum and use resources in the classroom is increasingly recognized as an important driver for the improvement of student learning outcomes (Bruns et al., 2019: 28). More importantly, our choice of focusing on teachers is also based on the observation that in both China and India, de jure and de facto policy development may not always be aligned (e.g. Tooley and Dixon, 2006;Li and Liu, 2020). For our purpose, the understanding of how CD programs work would nevertheless be incomplete without examining the experiences of the very recipients of such CD programs. The use of the teacher survey is thus justified as it allows for a structured exploration and comparison of teachers' on-theground experience of CD programs.
The teacher survey scrutinizes in-service training and career advancement as they are considered important teacher CD measures (Yan, 2019b). Instead of pre-service characteristics like teacher certification, level of education, and years of experience, it is the continuous updating of teachers' subject-matter knowledge, teaching skills, and ability to collaborate with and learn from peers that explains teachers' effectiveness in improving student learning (Aaronson et al., 2007;Jensen et al., 2016). Whereas in-service training is widely acknowledged as directly contributing to the accumulation of such competence, it should be noted that promotion or career advancement can play a role in enhancing CD by signaling and leveraging teachers whose individual capacity has been developed through CD programs as assets of the education system (Bautista et al., 2015). Teachers that receive promotion and recognition can provide in-service training or share valued experience and expertise with more junior colleagues. This creates a positive feedback loop as it develops expertise in content, pedagogy, management, or other aspects of school education.
The survey questionnaires asked about teachers' participation in different categories of in-service training (including those on textbook and syllabus, on pedagogy and teaching skills, on student and classroom management, on interaction with and involvement of parents, etc.). A similar set of questions asked about the promotion they had received, if any. Due to the lack of comparable objective indicators of student learning outcomes, effectiveness of CD measures is reflected through the level of teacher satisfaction on a Likert scale from 1 (most dissatisfied) to 5 (most satisfied). 3 The respondents were asked in both the questionnaire and in immediate follow-up interviews to further specify the reasons for their (dis)satisfaction. This combination of data-collection methods enables us to capture a rich and in-depth account of the experience and perception of teachers and helps us better understand how the variations of policy capacity may relate to perceived effectiveness. For instance, the account of whether and how they are consulted regarding CD needs and preferences helps inform analytical capacity of the CD providers. Comments and perceptions on the frequency, quality, and relevance of the CD programs, on the other hand, can give a more nuanced account of operational capacity. Lastly, whether and to what extent different stakeholders' interests are aligned and their actions are coordinated helps shed light on the political dimension of policy capacity.
While the lack of access to the full range of policymakers makes it practically difficult to reveal the entire process of designing and implementing CD programs in greater depth, we endeavored to compensate for this limitation by supplementing the teacher survey with a school/principal survey (following a largely similar procedure) whenever applicable, expert consultations and interviews with non-governmental organization (NGO) workers and government officials (see Table 1), and relevant policy documents and secondary data. These supplementary methods help pick up aspects that are not adequately covered in the teacher survey, such as the perspectives of the designing and delivering end of CD programs (government officials, NGO workers, and experts) or the budgetary information that teachers would have limited knowledge of, so as to generate a more balanced and comprehensive picture of policy capacity in CD programs.
The two districts of Fengtai and North Delhi were chosen by process of systematic elimination of incomparable districts within Beijing and Delhi, in which all middle schools were visited. From each participating school, teachers with varying years of teaching experience are invited to fill in the survey so as to check whether and how teachers at different stages of experience may have different CD needs and expectations. Participation consent was verbally acquired from heads of the schools and teachers before the start of the anonymous survey.
All respondents of the Beijing survey are regular teachers as school teachers are all regular employees in Beijing's government school system. In Delhi, guest teachers constitute a separate category alongside regular teachers employed as civil servants. Unlike their civil-servant counterparts, the contracts of guest teachers are renewable subject to attendance, on which their daily wages are based. 4 As the research aims to generate a comprehensive picture of CD in both countries, there is no ex ante rationale to exclude guest teachers in the survey. Accordingly, there are 150 respondents from all 33 government middle schools in North Delhi and 80 respondents from 22 out of the 24 schools in Fengtai, Beijing (see Table 2).

Findings
In-service training is regular and structured, while promotion structure is also clearly delineated, in both cities. Yet, beyond this bottom line, teacher CD practices vary substantially between government middle schools in Beijing and Delhi (see Table 3), not least in terms of both structures and formats, as elaborated in the following.

Variation in CD for teachers
Beijing has a comprehensive training provisions structure in which all levels of education bureaucracy are involved, with a certain degree of division of labor.
Yet, from the perspective of teachers, the training received by a vast majority of the survey respondents comes from schools and districts. In contrast, Delhi's training provision is centralized at the state level. As reflected in our survey findings, the most common and frequently mentioned training provider is the state education authorities, while the role of district and school providers is negligible (see Figure 2).
The format of training provision is also quite different between the two sites. Programs and sessions organized by the district and school providers in Beijing usually run throughout the semester. For instance, training on academic contents and pedagogy is held on a regular (often weekly) basis, with fixed and designated time slots. In Delhi, the state-dominated training is mostly concentrated during the (summer) break. Training within the semester is much rarer and ad hoc. Concentrated training during the vacations is mentioned by respondents in Beijing as well, though this is far from being the dominant format.
Beyond in-service training, the career advancement or promotion systems of the two regions have the starkest difference in terms of the promotion structure. In Delhi, promotion of (regular) teachers follows a "vertical" career path, in which teachers are shifted from teaching primary schools "up" to teaching secondary and ultimately senior secondary levels (see Figure 3). Teacher promotion in Beijing in terms of professional cadres, by contrast, can be described as a "horizontal path" in which primary and secondary schools are treated as separate structures. Apart from this most traditional sense of career advancement, there are also promotions within each cadre level, which are further supplemented by other arrangements of honoring teaching excellence or intermediate leadership positions such as "grade head" (nian ji zu zhang) or "director of political education" (zheng jiao zhu ren). Whereas within-cadre-level promotion with monetary increase may find some similarities in the Modified Assured Career Progression (MACP) (Iftikhar, 2018), promotion in the form of honorary titles and intermediate leadership positions, common not only in Beijing/China, but also in other Asian contexts such as Singapore (e.g. Bautista et al., 2015), was absent in Delhi during the time of the survey.
As a result of the variations in CD practices between the two regions, significant differences are observed in the level of satisfaction of teachers as regards the CD initiatives they receive. On average, satisfaction with in-service training is 3.2 (out of 5) in the case of Delhi and 4.2 in Beijing. In terms of career structures, more than 43% of respondents in Delhi have received at least one promotion in their career thus far, while the figure for Beijing is 52.5%. On average, time since last promotion for those promoted in Beijing is less than three years (33 months) prior to filling in the questionnaire. Yet, for their counterparts in Delhi, the interval is more than 79 months for current trained graduate teachers (TGTs) and 49 months for current postgraduate teachers (PGTs). The average rates of satisfaction with the career advancement system in Delhi and Beijing are 3 and 3.8, respectively.

Variation in policy capacity
While governments in both Beijing and Delhi are seen as quite active in their CD endeavors, these efforts have shown varying degrees of effectiveness, especially from the perspective of their direct recipients. This points to the need to examine the processes involved in carrying out CD, as suggested in existing CD literature, and the importance of the variation in policy capacity in explaining the observed differences in CD effectiveness (as summarized in Table 4).
Analytical capacity. In delivering in-service training, both governments have demonstrated limited analytical capacity in directly soliciting the needs and preferences of teachers. In our survey, less than one-third of the respondents in Beijing and 15% in Delhi report having been consulted (see Table 2). Both Beijing and Delhi turn instead to experts for inputs about in-service training during meetings at the state level at the planning stage before the start of a training cycle. At the time of the fieldwork, experts in Delhi were mostly from within the state-level authorities or schools of education in universities, who may not be up to date with the issues and challenges at the school level. The respondents noted that the training delivered by these experts often does not reflect and match on-the-ground realities, such as larger class sizes, students with slower progress, or inadequate facilities for science experiments.
In Beijing, such a limitation is partially compensated for by the opportunities for informal exchange enabled by a more decentralized training structure. Interviews with experts involved in the planning of in-service training also mention broader consultation during the planning stage with expert teachers, lower-level institutions, and academics working at the school level. Notably, experts in academia, such as university professors, are quite embedded as they are often deployed as academic mentors to teacher trainees in some CD programs. Some professors interviewed serve as vice principals to help design teacher development for particular schools, allowing them to closely follow the needs of CD through direct interactions with schools and teachers. None of these practices was found in Delhi.
It is also common for consultations in Beijing to include expert teachers, who, unlike university professors as "outsiders," have accumulated expertise directly from receiving earlier CD interventions and applying the learning in their teaching practices. Their resultant professional excellence is not only recognized through the career advancement system (i.e. being promoted and earning the status of expert teachers), but also further fed back into the CD system, in which they are tapped as trainers for subsequent teacher training. These expert teachers also remain similarly exposed to opportunities for connecting with teachers as the trainees' "practice mentors" (in parallel with university professors serving as "academic mentors") (see, e.g., Beijing Municipal Committee of Education of Communist Party of China [CPC], 2012), thereby allowing them to better discern front-line CD needs.
Another crucial element of analytical capacity is the collection of information from education institutions below the state level. India's District Institutes of Education and Training (DIETs) attempted to "improve government responsiveness to local needs" through more decentralized training provision (Dyer, 2005). Yet, according to our interviews with both government officials and the principal of one DIET, DIETs in Delhi have very limited autonomy over their own CD programs, which have a much smaller scale and coverage. They are also quite marginalized in contributing to the needs analysis of the state-level CD programs. In contrast, not only are district-level authorities in Beijing argued to enjoy more autonomy over training provided at their levels, but they also serve as an important source of information for state-level planning.
It is thus unsurprising that 50% of respondents in Beijing agree that existing training arrangements already match their needs and expectations, even though much fewer of them are formally consulted on the matter. Yet, in Delhi, 96% of the respondents indicate that current training arrangements do not match their needs or expectations. Zooming in on the different categories/subjects of training, in both contexts, training on classroom management and parental involvement remains inadequate, despite the perceived demand, as compared with training on academic contents and pedagogy (see Appendix Figures 1 and 2).
Operational capacity. Deficits in operational capacity in Delhi, as compared to Beijing, are mainly reflected in its selective coverage and low frequency of inservice training. Over 60% of the 32 surveyed guest teachers have not received training of any type. Even among the regular teachers, a little over one-quarter of the respondents did not receive any training at the time of the survey, the majority of whom had worked at their schools for less than a year. Low frequency of training and its concentration during the breaks also implies that missing, cancelling, or exclusion from the training delays the teacher's CD as opportunities will only come at the next training cycle-most likely during the following school break. Not surprisingly, "training frequency being too low" is cited as one major reason for dissatisfaction by respondents in Delhi (see Appendix Figure 3).
Delhi's lower operational capacity for teacher CD may reflect the differences in resources made available for education in general. In 2013, annual government education spending per student at secondary schools amounted to over CNY35,000 (USD5200 in current prices) in Beijing, making it the highest among all provinces in China (OECD, 2016). No available spending data are available for Delhi disaggregated by level, but recent estimates put per student state spending in Delhi at Rs9691 (USD131 in current prices) (Bose et al., 2017) and per student federal government spending at Rs43,171 (USD580 in current prices) using 2015 data. 5 Budgetary comparisons could be biased by different spending priorities and local-central government dynamics but the same difference in operational capacity is observed when pupil-teacher ratio (PTR) is examined. In 2015, Delhi has a PTR of 30 for secondary schools while Beijing's PTR is at 8.41 for junior secondary schools. These observed differences serve to confirm the variations in operational capacity found by the teacher survey.
Apart from more comprehensive coverage, Beijing's higher operational capacity is also reflected in its more targeted delivery, such as the district-level training program for "weak schools." In that program, each school sends five to six teachers to attend the training, which is usually conducted by a municipally renowned expert. The program includes diagnostic feedback on teaching demonstration provided by this expert, observation of their teaching, and action-research components, in which trainees are required to write a proposal, have it followed up by the expert, complete the study under their guidance, and write an essay toward the end of the program. Overall, the existence of multiple providers means that the training offered could at least be quite diverse, if not entirely complementary or synergetic.
Political capacity. Both Beijing and Delhi exhibited strong political commitment to the CD of teachers: although the absolute number of budgetary inputs is shown to differ substantially, the trend of budgetary increase is observed in both cities. Political commitment is crucial to mobilizing state funding, particularly in Delhi, where education financing is partially decentralized. From 2010 to 2015, public expenditure on basic education in Beijing has seen the largest growth when compared with the rest of China (Zhang, 2018). In Delhi, the Aam Aadmi Party (AAP) government has also allocated the largest portion of budget (over 20%) to the sector ever since being elected in February 2015 (PRS Legislative Research, 2019). Whereas Beijing's commitment is consistent with the importance attached to basic education at the national level ever since the establishment of the People's Republic of China (Rao et al., 2013), teacher CD has been a crucial component of Delhi's more recent education reforms (Sharma, 2020). While there is no evidence that differences in frequency and other aspects of training between Beijing and Delhi are caused by differences in budgetary commitment, substantial variations exist with regard to their respective pursuits of engaging different stakeholders and facilitating interest alignment in CD programs.
Involvement of NGOs in CD provisioning is almost negligible in Beijing (as revealed by Figure 2). The presence of NGOs as in-service training providers is more visible in Delhi: training offered by smaller-scale NGOs was often the only training they received over the surveyed period. Larger-scale NGOs (e.g. Pratham and Creatnet) are, in contrast, more actively involved in the government's CD planning as either partners or external advisors of the government (Ahmad, 2019;Ghavri, 2019). This signifies civic participation that is largely absent in Beijing. Nevertheless, concerns are raised about the compatibility of the approaches of NGOs with the traditional, government-led education governance structure (Subramanian, 2019). Nor are the NGO-led interventions always welcomed by the teachers, especially within the government school system (Gupta and Ahmad, 2016). In that case, the CD system in Delhi is yet to find a way to effectively steer and channel diverse stakeholders toward interest alignment and synergy.
This emerging tension between NGOs as an important external contributor and teachers as the ultimate recipients of CD also exposes a worrying impediment to consolidating political capacity in delivering Delhi's CD efforts, which is the inadequate engagement of the teachers. The study-abroad programs initiated by the state government with Harvard or Cambridge (The Statesman, 2019), albeit eyeopening, mostly benefit the "chosen few." Our teacher survey points to a general lack of regular and institutional channels within Delhi's centralized, top-down CD delivery through which teachers' needs and preferences for CD can be understood and deliberated. This is especially the case for guest teachers, whose reported training reception is significantly lower than that of regular teachers. They are also systematically excluded in the career ladder (as illustrated on the left side of Figure 3). As reflected in our survey, such exclusion is indeed one major reason for dissatisfaction with the promotion system.
In the case of Beijing, its political commitment to CD has been translated into a better-coordinated system. Through its diverse training provision across all levels, stakeholders are able to contribute in a complementary manner. Yet, instead of leaving these diverse CD programs as disjointed fragments, the system in Beijing shows remarkable political capacity in weaving them together into a functioning whole, which is mainly achieved by the state playing a supplementary role in areas that local levels are unable to cover, while recognizing and encouraging the robust training provision for average teachers already covered by local levels.
In terms of engaging key stakeholders other than local government and the schools, Beijing exhibits a higher capacity to reach out to teachers, whose voices enjoy multiple entry points for being integrated into CD planning and delivery. As mentioned earlier, not only do these channels provide a partial remedy to the analytical capacity deficit of not having formal and designated practice to consult teachers' CD needs, preferences, and expectations, but these multiple interactions also accumulate to promote trust among diverse stakeholders.

Conclusion
How does policy capacity affect the effective delivery of CD in the basic education sector? Using an original teacher survey and supplementary methods, this article has shown how variations in analytical, operational, and political capacities can jointly account for the perceived differences of CD effectiveness through a comparative analysis of teacher CD interventions in Beijing and Delhi. Findings from this article reiterate the necessity for policymakers and administrators to be concerned with whether the government has the capacity to effectively carry out CD interventions. In particular, without understanding and catering to the needs of the targets, CD initiatives meant to be supportive are likely to be dissatisfying and disappointing, as revealed in the teacher perceptions in the cases of Beijing and Delhi.
As a starting point of this type of research, this article paves the way for a research agenda on comparative studies of policy capacity in the field of education policy and beyond. First, the comparison of CD effectiveness may be further enriched if cross-national objective measurement of effectiveness, especially reflecting student learning outcomes, can be available to triangulate with teacher perceptions. This possibility is increasingly realized with the emergence of international large-scale assessments, with the notable example of the Programme for International Student Assessment (PISA). Beijing has participated in the last two rounds of the PISA test, whereas Delhi has not represented India in any of the PISA tests so far. While this possibility is welcoming, it should also be taken with caution in light of its limitations and controversies (Steiner-Khamsi and Waldow, 2018).
Second, similar research instruments can be used to explore CD for principals, school management committee members, government officials, or other CD program designers and providers in order to gain a more comprehensive picture about policy capacity and CD efforts for individual actors. When later opportunity arises, a closer scrutiny of the policy capacity needed for teacher CD programs interpreted from their perspectives can be expected to complement and enrich the picture generated in this article. This may also allow us to track the change and development of individual capacity over time.
Third, the survey employed in this article provides only a snapshot, albeit comprehensive, of Beijing and Delhi. In light of the ongoing reforms in both cities (see, e.g., Accountability Initiative, 2019; Meng et al., 2016), another round of surveys will be helpful in tracking the evolution of CD-related policy capacity.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.