Abstract
To make a comprehensive policy change, actors often turn to the gradual path where they introduce small-scale changes hoping that their accumulation will meet their goal over time. Nonetheless, they often stop the transformative process before meeting their original goal. This paper argues that this can be explained by policy learning. When actors learn from reliable information that the accumulation of the small-scale changes does not meet their expectations, they stop the transformative process. At the same time, the policy is not illuminated due to feedback effects and beliefs by the majority of actors that the small-scale changes are beneficial.
Introduction
Actors wishing to promote a significant policy change often face many institutional, political, and cognitive obstacles (Baumgartner & Jones, 2010). One way to bypass these obstacles is to gradually meet their goal by introducing small-scale changes that, over time, lead to a shift from the status quo. This strategy has been used in many policy fields in the last 60 years (Streeck & Thelen, 2005; van der Heijden & Kuhlmann, 2017). Some prominent examples are the Republican party’s efforts to replace Medicare by gradually introducing private insurance plans and blocking efforts to update the existing structures (Hacker, 2004) or the gradual changes made in Social Security (Béland, 2007).
Nonetheless, when closely examining the various examples, one finds that often, while actors were able to shift the status quo, they still had not met their initial goal after many years. For instance, both Medicare and Social Security are still active. Although some of the transformative processes might still be in progress, others stagnated and became adaptive; no new initiatives have been introduced, no expansion of the existing instruments or strategies has taken place, and most of the attention has turned to adjust existing structures. All that even though actors did not change their original goals and were not forced to stop the transformation. This raises the question of why actors stop the transformative process before meeting their original goal.
So far, we do not have a sufficient answer to this question. For the most part, existing research on gradual transformative policy change focuses on the shift from the status quo that has taken place and its causes. However, it does not examine to what extent this shift has fulfilled actors’ initial goals. This is important because focusing only on the successes of the gradual transformative change can lead to a distorted valuation of this change strategy, seeing it as more successful than it is.
The few existing works that look beyond the shift from the status quo have pointed to the role a powerful opposition can have on the stagnation of the transformative process (Mandelkern & Koreh, 2018) or to instances when negative feedback causes actors to reverse the transformation (Shpaizman, 2017). This paper aims to shed light on another mechanism that could operate in dynamics unfolding over a long period—policy learning.
Learning occurs when there are policy anomalies, discrepancies between the expectations and the outcomes (Hall, 1993; May, 1992; Sabatier & Jenkins-Smith, 1993). In a gradual transformative change, actors have expectations about the policy’s immediate outcome and the accumulated effect, the estimated effect of the accumulation of small-scale changes over time. The nature of the gradual process makes it likely to experience policy anomalies regarding the accumulated effect. Since the information processing of actors is not proportional (Baumgartner & Jones, 2015), actors will pay attention to the anomalies only when the level of conflict is moderate and when reliable information accumulates (Jenkins-Smith et al., 2014). Then, policy learning will take place; actors will update their beliefs about the expected accumulated effect of the instrument. In other words, they will no longer believe that the accumulation of small-scale changes will meet their expectations. As a result, they will refrain from further transforming the policy. At the same time, the long time during which the process unfolded often created stakeholders who might make it difficult, if not impossible, to replace the existing policy (Pierson, 1996).
This paper examines the suggested mechanism by applying process-tracing to one case from U.S. social policy, Head Start. This program is a positive case where a gradual transformation ended without reaching the actors’ original goal with no outside coercion or change in the actors’ goals. As opposed to many other social policies in the U.S., partisan polarization, in this case, is relatively low. This makes it a good candidate for examining explanations beyond party polarization.
Gradual Transformative Policy Change
In the last 60 years, many policies in the world have undergone transformative changes. In many cases, these changes were not abrupt comprehensive reforms; rather, they were a result of an accumulation of small-scale changes, taking place over time and leading to a significant shift from the status quo (Hacker, 2004; Mahoney & Thelen, 2010; Streeck & Thelen, 2005). Actors often gradually promote their goal when facing significant barriers to doing so abruptly (e.g., through comprehensive legislative reform). Such barriers might include powerful veto players, institutional veto points, or status quo bias (Hacker et al., 2015; Mahoney & Thelen, 2010). Alternatively, actors have chosen the gradual path when much risk or uncertainty is involved, which can be better contained when the process is gradual.
Many works have demonstrated the existence and evolution of gradual transformative changes (Béland, 2007; Christensen & Grossi, 2022; Galvin & Hacker, 2020; Hacker, 2004; Mahoney & Thelen, 2010; Streeck & Thelen, 2005; van der Heijden & Kuhlmann, 2017). Although a gradual path seems an appropriate substitute for an abrupt one, so far, we do not know whether, over time, it can meet actors’ initial goals. This is because existing research focuses almost entirely on the shift from the status quo that has taken place and neglects the relations between the outcome of gradual transformative change and the actors’ initial goals.
This would not be important if most (or many) of the gradual changes eventually met the actors’ goals. However, when examining existing works, we find that, while in some cases, the gradual dynamic displaces existing structures or opens new opportunities for a more comprehensive reform (Falleti, 2010), in many other instances, this is not the case. For instance, the Republican party in the U.S. wished to introduce private savings accounts that would eventually lead to the privatization of Social Security (Béland, 2007; Hacker, 2004). Although these accounts have significantly expanded, they have not replaced Social Security (Kelly, 2016). This outcome is common in many other gradual transformative policy changes (van der Heijden & Kuhlmann, 2017).
As long as we focus on the successful shift from the status quo and ignore the end result of the process, we perceive gradual changes as more successful than they might be, gaining not only a partial but also a distorted understanding of them. To fully understand the potential and limitations of gradual transformative policy changes, we should look at the factors that enable shifting from the status quo and the obstacles gradual changes might experience in the long run.
Several recent works have looked at gradual transformative changes beyond the shift from the status quo. Mandelkern and Koreh (2018) have argued that gradual change can be stopped by powerful opposition pushing for reversal. Shpaizman (2017) found that when the gradual process experiences negative feedback (Weaver, 2010) undermining its benefits, that can be a subject for reversal. Although these works are important in advancing our understanding of gradual transformative changes, they do not explain many other instances in which the gradual process was not reversed but rather stagnated. Because a gradual transformative change is a long-term process, a possible explanation for its stagnation might be a change in the actors’ beliefs, namely policy learning.
Policy Learning
Learning is the process of updating beliefs about key components of policy, namely, beliefs about causal relations, the desired policy outcome, and the strategies and policy instruments used (Bennett & Howlett, 1992; Greener, 2001; Sabatier, 1988). Learning approaches in policy are based on the assumption that policymakers can learn from experience and modify present actions based on the outcomes of past actions (Bennett & Howlett, 1992). Learning takes place over a decade or more (Sabatier & Jenkins-Smith, 1993). It can be a deliberate activity when actors seek information to adjust the policy (Hall, 1993) or a stimulus for social and environmental pressures (Heclo, 2010). Learning occurs when individuals assimilate new information from their own experience or the experience of other actors, evidence-based analysis, and social interactions over time (Radaelli, 2009; Thunus & Schoenaers, 2017). It occurs at the individual and the collective level and in the interaction between the two (Heikkila & Gerlak, 2013).
Learning can have two outcomes, cognitive and behavioral. The cognitive outcome can be a change in the beliefs regarding policy instruments, processes, or causal relations within the policy system. The behavioral outcome can be a change in the policy settings, instruments, or goals (Gerlak & Heikkila, 2011; Heikkila & Gerlak, 2013). The outcome of learning can be “simple” (also termed instrumental learning), resulting in technical adjustments of the instrument or the strategy used, or deeper and more complex (societal learning), resulting in a revision of the policy objectives and the causal relations underlying the policy (Hall, 1993; May, 1992).
Learning takes place as a response to an accumulation of policy anomalies; discrepancies between the expectations of the policy (the policy beliefs) and the policy in practice (Hall, 1993). Anomalies may result from changes in the environment, new information, or experimentation that was not evident before. Anomalies can exist at different dimensions: the instrument settings (how instruments are implemented), the instruments themselves, pragmatic objectives, or broader societal goals. Learning occurs when policymakers pay attention to the anomalies and interpret them based on the existing information to assess and label problems in the existing policy (Wilder & Howlett, 2014, 2015).
Learning is not linear and Bayesian, so in the face of new information, policymakers do not immediately change their views or behavior (Nowlin, 2021). Policy anomalies are informational signals. They are not self-evident but instead constructed. As such, they may exist for a long time before attracting attention. There are various and ambiguous informational signals, but not all receive attention. Those that do are processed disproportionality, with some receiving attention and others ignored (Baumgartner & Jones, 2015; Holcomb et al., 2009). This is because of individuals’ and organizations’ bounded rationality (Simon, 1972). In addition, policymakers do not so easily admit mistakes and change their preferences (Moyson et al., 2017). They tend to interpret new information based on their previous convictions and various heuristics, and they resist change until they are either forced to acknowledge it or until the accumulated evidence can no longer be ignored (Dunlop & Radaelli, 2017; Jones & Baumgartner, 2005). Lastly, the anomalies are not always visible; in some cases, it takes long before actors recognize them (Sabatier, 1988).
Several conditions are more conducive to learning. Learning is more likely when the information is credible, that is, scientifically based, and has little uncertainty (Nowlin, 2021). Moreover, a single study or report hardly ever affects the beliefs of the political actors. Therefore, the information should also be accumulated over time to be harder to ignore (Weible, 2008). Furthermore, policy learning is more likely when the level of conflict between the actors is moderate. This occurs when the issue is not politically polarizing and not entirely technical. When the level of conflict is low, it is difficult to draw actors’ attention to the new information (see, for instance, Baumgartner et al., 2009). On the other hand, when the level of conflict is high, each side defends its original positions and does not wish to pay attention to new information (Bundi & Trein, 2022; Nohrstedt et al., 2017).
Gradual Change and Policy Learning
Learning can occur when there are discrepancies between the actors’ expectations from the policy and the policy in practice. When actors choose to promote their goal gradually, they have two types of expectations from the process: first, the effect of each small-scale change; “expected immediate effect,” and second, the effect of the accumulation of the small-scale changes over time (the actors’ initial goal; ‘expected accumulated effect.’) “expected accumulated effect.” For instance, when the Republicans introduced personal retirement saving accounts, they expected the accounts to act as an acceptable alternative for pension saving and also anticipated that the expansion of the accounts would, over time, undermine and ultimately replace Social Security (Béland, 2007; Hacker, 2004). The expected accumulated effect may be on the level of the policy instruments, when the actors expect that the accumulation will displace an existing service, or on the level of broader societal goals when the actors expect that the accumulation will have some broader effect on society. In some cases, both types of expectations are present. In a gradual transformative process, actors can learn about the accumulated effect, the immediate effect, or both.
The characteristics of gradual transformative changes create a fruitful ground for policy anomalies regarding the relations between the expected accumulative effect and the effect in practice. Gradual transformative change is usually not carefully planned. It evolves in an ad-hoc nature by seizing institutional and political opportunities (Christensen & Grossi, 2022; Hacker et al., 2015; Koreh et al., 2019; Mahoney & Thelen, 2010; Streeck & Thelen, 2005). In some cases, the small-scale changes are experimental ones whose effect is not fully known, especially once expanded. This is especially true for various social intervention programs involving complex mechanisms and uncertainty. Hence, actors might unintentionally select an instrument or strategy that cannot meet the expected accumulated effect.
In addition, gradual change evolves over a long period during which changes in the policy environment might occur, influencing the accumulated effect. Moreover, due to the long-time span, actors might choose a suboptimal instrument/strategy which enjoys broad support, hoping that either, over time, they will be able to improve the instrument’s capacity or that the blame will be targeted on those who come after them (Palier, 2007).
As mentioned earlier, there is often a delay between the time the information is received and the time the update of the beliefs takes place (Jones & Baumgartner, 2005). In the case of gradual transformative change, the difficulty is even greater because there is much uncertainty involved when assessing the full potential effect of accumulating several instruments or expanding a certain instrument. Therefore, it is likely that, at first, actors will interpret the anomalies as temporary or manageable and only later as permanent.
The long time span of the gradual change can be conducive to learning. The long period enables the accumulation of reliable information on policy anomalies. In addition, over time, the political environment often changes, which, in turn, can affect the level of conflict on the policy issue. Thus, even if, when the policy was adopted, the level of conflict was low, with time, this might change because of changes in the actors and their beliefs or in the power of various groups. For instance, research on American politics found that issues once seen as bipartisan (e.g., education) became more partisan over time (Jones et al., 2003). This was, among other reasons, due to changes in legislatures I power and the power of various interest groups (Hacker & Pierson, 2008). The accumulation of reliable information and the change in the level of conflict can lead to policy learning. Actors can learn about the limitations of the gradual path or the selected strategy (Christensen & Grossi, 2022; Galvin & Hacker, 2020; Koreh et al., 2019) or the inappropriateness of the tools to meet the goal.
Gradual accumulation is often not inertial. Each small-scale change is deliberate, often requiring coalition building or other measures (Béland, 2007; Hacker, 2004). Therefore, when a majority of policymakers start questioning whether the accumulation of small-scale changes can meet their expectations, the change in the belief system can lead to a change in behavior. In other words, actors might be reluctant to continue and expand the instrument because of the deliberate effort it requires and given their limited resources such as agenda capacity, budgetary constraints, political capital, and others.
At the same time, gradual transformative change might create a feedback effect that makes its reversal difficult, if not impossible. During the period the policy evolved, new stakeholders have likely been created, or the policy may have gained public and political support (Béland, 2010; Pierson, 1996). The feedback effect creates support for the maintenance of the policy instruments even if it does not meet the expected accumulated effect (Moore & Jordan, 2020). Moreover, while the accumulated effect is not what was expected, the outcome might meet part of the actors’ expectations. In such a case, actors will wish to maintain what has already been achieved, performing minor adjustments when needed. Lastly, actors are more likely to abolish a policy in place for a long period when new ideas on how to handle the situation are available and when the various interest groups are willing to adopt them and push them into the institutional framework. If the interest groups do not wish to push for new ideas due to the existing feedback effect or when new ideas are unavailable, the result will be continuing adjustments of existing policy tools (Blyth, 2013; Oliver & Pemberton, 2004). Consequently, despite the policy anomalies, replacement of the existing policy evolving through a gradual process will be less likely, unless negative feedback or a powerful opposition is pushing for it (Mandelkern & Koreh, 2018; Shpaizman, 2017). The result will be stagnation in the transformative process and technical adjustments of the existing policy tools.
To sum up, a gradual transformative process might stagnate before meeting actors’ initial goal when learning on the accumulative effect of the policy takes place. This can happen when there are anomalies at the accumulated level and reliable information about these anomalies accumulates, and the level of conflict is moderate. In that case, actors will pay attention to the anomalies and change their beliefs about the ability of the gradual steps to meet their original goal. At the same time, while the process stagnates, the existing policy tools will remain in place and be adjusted as long as there is a feedback effect preventing policy replacement or when the actors will continue to believe that the immediate effect of the policy is positive. Figure 1 summarizes the suggested mechanism.
Gradual transformative policy change and policy learning.
Methodology and Data
The present analysis illustrates the suggested mechanism using process tracing analysis (Beach & Pedersen, 2013) of one case in U.S. social policy: Head Start from its establishment in 1965 until 2020. This program was chosen because it represents a positive case of gradual transformative change, which became adaptive before meeting actors’ initial goals.
The data are taken from a variety of sources. In each source, all the relevant observations during the period under examination were studied: U.S. Federal budgets, the president’s budget proposals, reports of the Congressional Research Service (CRS), Administration for Children and Family (ACF) reports and budgets, Head Start administration statistics, Congressional hearings addressing Head Start (32 hearings), State of the Union Addresses, Democratic and Republican party platforms and secondary analysis of existing research. All budget figures are presented in real values. In addition, a data set that includes all the bills introduced regarding the program has been constructed (105 bills). The data set is based on Adler and Wilkerson (2012). Each bill was coded to indicate whether it suggests continuing the transformation (expanding the program) or making minor adjustments (improving the quality/accountability).
Table 1 summarizes the predicted evidence for each causal condition in the analysis and the data used for each prediction. The table is based on Beach and Pedersen (2013, p. 169). For each condition, one or more of the stated predictions are used. Because the analysis aims at examining the mechanism of learning, the absence of negative feedback and the absence of a powerful opposition are defined as scope conditions. The collected evidence has also been used to examine the relevance of two alternative explanations: that the stagnation is a result of party polarization preventing policy expansion and that the stagnation is a result of a change in the actors’ goals.
Head Start: Predicted Evidence and Data used to Measure Predictions.
Head Start
Gradual Transformative Change and Expectations
Head Start is the first federally funded preschool program (Morgan, 2001). It started in 1965 as an 8-week summer program for children from disadvantaged backgrounds. The immediate aim of the program was to help preschool children be ready for school by providing comprehensive services (social, emotional, educational, nutritional, and health). The program was part of President Johnson’s war on poverty. From its establishment, it had the broader societal goal of eliminating the causes of poverty. Specifically, the program’s initiators believed that Head Start could significantly narrow educational gaps between children from different backgrounds by closing the gaps when the children enter school (BOAI, 2017b; Karch, 2017; Westinghouse Learning Corporation, 1969). President Johnson believed that to narrow educational gaps (which are some of the causes of poverty), Head Start should eventually reach all eligible children (Johnson, 1967).
Over time Head Start significantly expanded. In 1966, it became a year-round program. In 1981, The Head Start Act was passed, making Head Start an individual discretionary program. In 1994, Early Head Start was added to address infants and toddlers under three (ECLKC, 2017). In addition, the number of funded slots increased from 561,000 in 1965 to 852,501 in 2020. The allocated funding has also increased from ~792$ million in 1965 (in 2020 dollars) to ~10.6$ billion in 2020. The funding per child also increased, from ~1,400$ (in 2020 dollars) in 1965 to ~12,000$ in 2020 (see Figures 2 and 3). Today Head Start is the main federal preschool program. Its budget is 17% of the ACF budget, and since its establishment, it has served more than 22 million children (ACF, 2021, 2022; CRS, 2014).
Head Start funding.
Head Start funded slots.
The expansion of Head Start was possible because of broad public and political support based on the belief that the program helps eliminate the gaps between children when they enter school, and by doing so, it contributes to their later success in life (Morgan, 2001). Correspondingly, a bipartisan consensus was created that Head Start should be expanded to reach its full potential. Policymakers and politicians from both parties termed Head Start a model, a success story, and even compared it to a vaccine that, once found, should be used to cover as many people as possible (Clinton, 1993; Committee on Education & The Workforce 1998a, 2003). Moreover, even when other social programs were significantly cut, or the party majority changed, a consensus around Head Start continued (Bush, 1989, 2001).
Policy anomalies and information
Child development research was not very broad when Head Start was introduced. As a result, Head Start was designed based on the intuition and hunches of its developers, and there was much uncertainty regarding its outcomes and impact (Zigler & Muenchow, 1994). The first study to evaluate the impact of Head Start was conducted in 1968. The study found very little effect of the program (Westinghouse Learning Corporation, 1969). This study was widely criticized for its methodology, and its findings were rejected by policymakers, parents, and interest groups (Karch, 2017; Morgan, 2001).
From then and until the end of the 1990s, there were no formal evaluations of the program (although various studies of the program were conducted). Policymakers based their assessments on state reports and the experience of individuals usually expressing high satisfaction with Head Start. Based on this information, they constantly stated that the program was very successful in helping children to be ready for school and beyond and, therefore, should be expanded (Clinton, 1993; Committee on Education & Labor, 1985; Committee on Labor & Human Resources, 1988).
The first signs of policy anomalies were presented to members of Congress during the deliberation on the 1998 Head Start reauthorization act. State administrators and local providers testified that, based on their experience, children who finish Head Start continue to lag behind their peers in literacy and math. In addition, some suggested that the effect of Head Start fades out so that Head Start graduates do not succeed in school as well as their peers. Policymakers interpreted these shortcomings as manageable. They believed that by moderately improving Head Start’s academic components, the program’s effect would improve, and in the long run, educational gaps would narrow. Correspondingly, the goals for the 1998 reauthorization process were to improve and expand Head Start (Committee on Education & The Workforce, 1998a, 1998b, 1998c). In addition, the bills introduced during that period continued to suggest expanding the program and even making it an entitlement. Although some conservative members of Congress argued that before further expanding the program, it should first demonstrate that it could fulfill its expectations, these voices remained marginal (Committee on Education & The Workforce, 1998b).
In 2003 Congress began once again deliberating Head Start reauthorization. This time policymakers had more information regarding Head Start’s impact. From 1997 the ACF began collecting data on Head Start in the form of longitudinal surveys of Head Start participants, termed FACES (The Head Start Family and Child Experience Surveys). All surveys systematically confirmed the actors’ beliefs about Head Start’s immediate effect, that is, that Head Start children perform better than children who did not participate in Head Start. At the same time, they pointed at the policy anomalies at the accumulated level, that is, despite the expansion of Head Start, the gap between children from different backgrounds was not eliminated as they entered school, so the disadvantaged children continued to lag (ACF, 2017a). When the survey results were presented to Congress, more members of Congress began questioning whether Head Start could meet their expectations of narrowing educational gaps.
Despite the investment in the program over its nearly 40-year history, a significant question has been raised about the effectiveness of Head Start. Study after study has documented how children who enter Head Start are better off when they leave. That information is encouraging and is appropriately brought up in this hearing, but a troubling statistic that has accompanied many of these studies is that children leaving Head Start continue to lag behind their peers who come from more advantageous circumstances. (Senator Mike Enzi, R-WY, Committee on Health, Education, Labor & Pensions, 2003)
Moreover, members of Congress saw that Head Start’s impact remained mostly unchanged despite the changes made in 1998 (Committee on Education & Labor, 2007a; Committee on Health, Education, Labor, & Pensions, 2003). As FACES data accumulated, the idea of expanding Head Start was almost entirely removed from the agenda. The reauthorization hearings from 2003 for instance, did not include expansion as one of the goals of the process as before. Furthermore, from 2003 until 2020, only six bills suggested expanding the program. The rest addressed the program’s various components, such as staff salary or the program’s hours. Almost all presidents’ budget proposals (except the 2016FY budget) since 2003 have not suggested expanding but rather maintaining the existing level of enrollment (GPO, 2003–2020). Occasionally, some members of Congress addressed the need to expand the program. Nonetheless, more often than not, it seemed more like a symbolic gesture, part of their opening statements. Because they did not continue to address this issue during the rest of the hearing, nor did they suggest any operational solutions on the subject (Committee on Education & The Workforce, 2003, 2005; Committee on Health, Education, Labor & Pension, 2003).
In 2009 the ACF published the first long impact study results, comparing children who participated in Head Start to children who did not participate even though they were eligible to. The study examined the children’s achievements over time from 2000. It found that the program’s effect on children’s educational achievements decreased by the end of first grade. A follow-up study conducted in 2012 found that the effect almost faded out completely by the end of third grade (ACF, 2010, 2012). These results received great attention from policymakers and the media (Besharov & Call, 2009; Strauss, 2013). The studies were most likely the final straw leading members of Congress to give up on their beliefs that Head Start could narrow educational gaps and, as such, should be further expanded. From 2009 we find no more expressions arguing that Head Start could help children succeed in school and beyond, and only two bills were introduced suggesting expanding the program.
Moderate level of conflict
Since 2003, there was an increase in the level of conflict over Head Start. Although the program remained consensual since no actor suggested abolishing it and all agreed that it was beneficial, there was conflict on how it should be funded. The administration pushed for funding through block grants delivered to the states, and Democratic representatives and Head Start advocates opposed this. The administration argued that block grants would enable more flexibility, improving the program’s outcomes. The opposition to this idea argued that block grants would eventually dismantle the program. While the suggestion to fund Head Start through block grants was not new, from 2003, it was placed high on the agenda (Pasachoff, 2006). The proposal to block grant Head Start was part of a broader conservative agenda of reforming the federal aid system, which was highly conflictual (Conlan, 1984; Rocco, 2015). As such, it increased the level of conflict over Head Start (Committee on Education & The Workforce, 2003, 2005; Committee on Health, Education, Labor, & Pensions, 2003).
Stagnation in the transformation and shift to adaptation
At the same time as the level of conflict over Head Start increased (2003 reauthorization), and along with the accumulation of information, the expansion of Head Start in terms of the funded slots stagnated and even decreased a bit from 905,235 in 2001 to 852,501 in 2020 (ACF, 2021) (see Figure 3). This is despite the addition of Early Head Start, which increased the number of eligible children (Subcommittee on Early Childhood, Elementary & Secondary Education, 2017). In addition, the budget appropriations from 2003 and the 2007 reauthorization of the program were explicitly designed to maintain the existing number of children enrolled. Although until 2017, the presidential budget proposals suggested increasing the allocated funding, this increase aimed to expand the program’s hours or the teacher salaries and not the number of funded slots (White House, 2016). Correspondingly, while until 2018, there was an increase in funding, this increase was not used to expand the number of funded slots but rather adjust the existing structures (CRS, 2006, 2014; Committee on Education & Labor, 2007b; Committee on Education & The Workforce, 2015; Subcommittee on Early Childhood, Elementary & Secondary Education, 2017). Since 2018 the president no longer suggested increasing the program’s funding.
Head Start stopped being expanded before reaching all eligible children as had been hoped for. By 2017, it served about 42%, and Early Head Start served only about 4% of income-eligible children. Policymakers were aware of these figures (Committee on Education & The Workforce, 2003, 2005; Subcommittee on Early Childhood, Elementary & Secondary Education, 2017).
Feedback Effect and Beliefs on the Program’s Positive Effect
While actors began questioning and later changing their beliefs about Head Start’s accumulated effect, no change in the beliefs at the immediate level took place, and actors continued seeing Head Start as a beneficial program that assists with school readiness for underprivileged children (Committee on Education & Labor, 2007b; Committee on Education & The Workforce, 2015; Subcommittee on Early Childhood, Elementary & Secondary Education, 2017). In addition, by the end of the period under examination, more than 1,600 nonprofit organizations were operating the program (CRS, 2021). These organizations were interested in the program not closing down (Karch, 2017). This can explain the maintenance of the program.
Head Start stagnation was most likely not a result of party polarization. The bipartisan support for the program did not change after 2003. Although the stagnation began under a Republican government, which cut many social programs, the Head Start budget was not cut; it increased just enough to maintain the existing services (Bush, 2002). The Head Start 2007 reauthorization act was introduced by Democratic members H.R. 1429-110th Congress (2007–2008), but passed with bipartisan support in both houses (95-0 in the Senate and 381-36 in the House) and was signed by a Republican president. Lastly, since 2003 both parties introduced only six bills suggesting expanding Head Start. As a result, there were very few opportunities to block initiatives to expand the program.
The stagnation was also not a result of a change in the actors’ goals. The learning centered on the beliefs and on the policy tools (the extent to which Head Start can narrow educational gaps), and not the policy goal, as the actors continued to express their strong belief in preschool education as a key to narrowing educational gaps (Clinton, 2016; Committee on Education & The Workforce, 2015). On the contrary, the idea that narrowing educational gaps through preschool education is the key to reducing poverty became stronger, as the main goal of the U.S. preschool policy became school readiness (CRS, 2018; Karch, 2017). This is also evident in the saliency of early childhood education in the 2016 election campaign for both parties (Clinton, 2016; Trump, 2016). The summary of the findings can be seen in Table 2.
Head Start from transformation to stagnation and adjustment.
Conclusion
Actors wishing to make a significant change in the status quo often choose a gradual path. However, although this choice has many advantages, actors might decide to stop the policy transformation although they did not meet their initial goal. This paper aimed to provide one explanation for this outcome. It has been argued that the nature of the gradual change process makes it more likely to experience discrepancies between the expected accumulated effect and the effect in practice. At the same time, a long-term gradual process is also conducive to policy learning. Over time, reliable information on the policy anomalies can accumulate, and the level of conflict can change, which can turn actors’ attention to the policy anomalies. As a result, actors can change their beliefs on the ability of the gradual path to meet their original goal. At the same time, the long-term process often creates stakeholders who oppose the reversal of the process or believe that the policy’s immediate outcome is beneficial. Correspondingly actors will stop the transformative process but will not replace the existing policy.
This argument has been examined in the case of Head Start. In this case, actors saw the program as a type of remedy that should be applied to as many children as possible to narrow educational gaps and reduce the causes of poverty. The policy anomalies were evident for a long time. However, policymakers either ignored them or believed that they could be managed. When reliable information on the policy anomalies like formal surveys and evaluations accumulated and the level of conflict over the program increased, policymakers paid attention to the information. This, in turn, made them change their beliefs that Head Start could narrow the educational gaps between children from different backgrounds. As a result, the program’s expansion stagnated even though it did not reach all eligible children. Since policymakers continued to believe that Head Start benefits its participants and given the strong support for the program, they did not replace it but continued adjusting it.
This paper suggests that the gradual transformative path cannot be a sufficient substitute for an abrupt and comprehensive change. Actors who wish to make a significant reform and cannot do so abruptly should consider that choosing the gradual path will most likely take them only part of the way. They can use it to legitimize some intervention or get a foot in the door. However, if they wish to achieve more comprehensive reform, they probably should keep looking for opportunities for comprehensive legislative reform. This was, for instance, the case with the Health Care Reform in Brazil (Falleti, 2010).
These findings do not aim to undermine gradual transformative changes’ significance but rather to point at their limitations. So far, we have known the conditions under which actors choose the gradual path; by knowing their limitations, we can examine when and under which conditions they return to the abrupt strategy. This will provide a better understanding of policy and institutional dynamics.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
