82661 Competitive Industries Note: Flexible Delivery: Organizations that Can Fail and Still Succeed Industrial policy needs to be flexible As Albert Hirschman showed, long-term development often requires continuously finding partial solutions and switching between approaches, as much as finding an “optimal” policy mix (Hirschman, 1970). This is nowhere more true than in industrial policy, which tackles complex problems, must interact with rapidly changing global markets, and requires difficult and uncertain choices under conditions of uncertainty. As importantly, there are many different forms of industrial policy, each appropriate to a different local context. Many of the errors of the past have come from deploying a top-down approach using instruments that were poorly fitted to the local political economy. That poor fit manifested itself in the details of the design All of this must be grounded in institutions. Policy and implementation of policies, but getting those details programs are run by agencies, so for the programs to be right requires iterative change and learning (Rodrik, 2008). flexible, the agencies must be flexible, whether Ministries Iterative change, in turn, requires flexibility: the ability to try or small “reform teams” (Criscuolo and Palmade, 2008). In an approach, understand if it is working or not, and fine- a bureaucracy, however, flexibility is easier said than done tune or discard it. (Wilson, 1989). Many agencies try to be flexible, but fail, either becoming rigid de facto or exploiting the discretion Table of Contents required by flexibility, and shirking or being captured. DARPA: Extreme Flexibility in Action. . . . . . . . . . . . . . . . . . . . . . . . . .1 There are, though, some agencies that have managed Flexibility in the History of to achieve flexible delivery, and been instrumental in Industrial Policy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 transforming countries and industries. One of the most The Theory of Flexibility and Risk. . . . . . . . . . . . . . . . . . . . . . . . . . . .3 striking is DARPA. about the authors DARPA: extreme flexibility Luke Simon Jordan in action Luke Jordan is a Private Sector Development Specialist with the Competitive Industries Practice, based in New Delhi. He works on The Defense Advanced Research Projects Agency manufacturing in India and mining based growth in Afghanistan. (DARPA), part of the U.S. Department of Defense, is one of the most flexible of all public organizations. It can claim Katerina Koinis a key role in the birth of the internet, the global positioning Katerina Koinis is a Private Sector Development Consultant with the Competitive Industries Practice, based in Washington DC. system, and stealth. She is a core member of the Practice’s knowledge program on the implemention of industrial policy. One of DARPA’s most important features is its clearly defined relationship to its sole client, the U.S. military, and its mission, to avoid technological surprise. The focus and Financial & Private Sector Development — October 2012 1 Comptetitive Industries Note | October 2013 • Issue 3 1 clarity of a long-term mission has created discipline and active talking. PMs, in this respect, are facilitators. They liberated short-term programs from incrementalism. take an active role in discussing with the performers and deciding on results. The ecosystem of institutions within which DARPA operates is also vital to the agency’s effectiveness. DARPA It is interesting to note what DARPA does not do: peer neither conducts research itself nor implements any of its review; results measurement; “killing” programs; long-term programs. It funds researchers and connects them and career development; formal performance incentives; or their output to customers. Without them, DARPA would strict controls against revolving doors, capture or collusion. not be able to operate, as a venture capital fund cannot Given this, DARPA might face risks to its effectiveness function without a deal flow. due to informal collusion and capture. This is a substantial DARPA also relies on a collection of outside contractors and present threat to the agency. Besides the rigorous for much of its own activities. This allows it to keep its reviews, continuous flow of information, and the hard permanent staff small, only 100-150 program managers discipline of the long-term mission, one of the key ways (PMs). this risk is mitigated is the informal practice of never relying on a single institution, person or organization for too long. These PMs arrive from academia, government and Instead, DARPA continually seeks to seed capabilities in industry. One of DARPA’s most striking organizational new hosts, so that it only faces a monopoly for as short as features is their 4-year term limits. This screens personality possible. type: PMs have to be innate risk-takers, believing that in 4 years they will achieve a breakthrough that will be In all this, what is vital is not the ability to fail, but the beneficial both to the agency as well as to their later career. ability to admit failure and redirect resources away from Moreover, the term-limit strongly supports flexibility: those it. DARPA’s reputation and the separation of programs who might resist a change in direction leave within a few and approaches – where even if an approach may fail, a years. program can still persist – are among the key features that help overcome obstacles leading to the admission of failure. DARPA’s basic decision-making architecture combines rigor in reviews with autonomy between them. Programs must be approved by the Office Director (OD) and the Flexibility in the history DARPA Director, following the “Heilmeier Catechism” of industrial policy (Table 1). After and between reviews, however, PMs have Several examples can also be drawn directly from industrial almost complete autonomy in writing bid documents and policy. Perhaps the most famous is the Ministry of Trade contracts to implement approaches. To decide to tackle a and Industry (MITI) in Japan. In facilitating Japan’s rise, it problem the bar is set very high, seeking to avoid mistaken would often try several different approaches before finding programs, but once underway, the bias is towards not one that worked. In the auto industry it tried to induce missing opportunities to succeed. The first means high ‘rationalization’ through mergers between carmakers, approval thresholds; the second means autonomy in action a move successfully – and fortunately – resisted by the (Sah and Stiglitz, 1986). industry. It then tweaked the approach, and deployed it DARPA contracts are informally known to set performance with auto components industry, with far more successful goals that are impossible to achieve. This triggers results. continuous problem-solving discussions, and creates a Likewise, through the 1950s and 1960s it tried an array of continual justification for PMs to rework contracts, should approaches to facilitate Japan’s exit from coal mining. It they wish to. The agency is then sometimes described finally achieved this through training for workers, policies to as “spending all its time talking” – but this is purposeful, 2 Comptetitive Industries Comptetitive Note Industries | October Note 2013 | October • Issue 2013 3 3 • Issue 2 Table 1: The Heilmeier Catechism A theory of flexibility 1 What are you trying to do? Articulate your objectives using absolutely and risk no jargon. The organizations that succeeded in delivering flexibly 2 How is it done today, and what are the limits of current practice? managed – at least for a time – to solve three interlocking 3 What’s new in your approach and why do you think it will be suc- problems: creating enough discipline to allow discretion cessful? without shirking and indiscipline; managing the political 4 Who cares? If you’re successful, what difference will it make? costs of failing while keeping enough support to be 5 What are the risks and the payoffs? effective; and deciding, under uncertainty, when a 6 How much will it cost? How long will it take? program, policy or investment is a failure, or is just facing 7 What are the “midterm exams” and “final exams” to check for suc- teething trouble. cess? These can be summarized as governing discretion; managing exit costs; and making decisions under enable shipping, investment in foreign mines, and more. uncertainty. Table 2 provides a summary of ways in which Throughout, MITI kept the number of key officials small DARPA and some of the agencies described above have – under 100 – and rarely, if ever, sought to wield direct solved them. Three key strategies stand out: control. 1. Set an ambitious goal, unambiguous and easy to A similar story from East Asia can be told about the Blue observe, roughly a decade in the future, and tie the House Secretariat in Korea. With a small staff mediating agency’s existence to it – this hard wall of future between Ministries and bureaus, it oversaw a decade discipline then allows flexibility now of approaches to fostering the emergence of a globally competitive steel company and automakers, trying and 2. Distinguish between “programs” and “approaches”: jettisoning one consortium and policy package after set a high bar for approving programs, but make another until it found solutions (Kim, 2011). approaches easy to try and to exit – concentrate failure and exit in approaches In France there was the Commissariat Général du Plan (CGP), founded by Jean Monnet in 1946. Its mission 3. Make officials of the agency accountable above all for and nature changed enormously as it sought to facilitate surfacing and using information, using difficult goals France’s recovery from World War II. The first years as a tool of people and information management rather focused on heavy industry and agriculture, the subsequent than a substitute years on greater ‘modernization’. Throughout, it stayed Finally, these agencies stay small – almost never above small and used a long list of supporting agencies and 100 people – and in recruiting they bias towards practice, councils to implement flexibly (Cazes, 1990). especially engineers and officials, over theory. It is clear that much of this in practice is a tall order. In many contexts it will be impossible. With sufficient care, and hard choices in framing their mission, they can be introduced in unlikely places, and when they are, they can change history. Comptetitive Industries Comptetitive Note || October Industries Note October 2013 2013 • • Issue 3 Issue 3 3 3 Table 2: Fourteen Contributing Factors in Flexible Delivery Structural characteristic 1 A threat made politically salient to a broad section of the elite 2 A clear and simple mission whose achievement will answer the threat, and whose failure will make its realization much more likely 3 Ambition and simplicity of measurement in defining the goal, more so than attribution (removing the ability to “fudge” success is more important than being able to attribute it) 4 Strong but delayed incentives, collective (the end of the agency if it fails) and individual (career-making or career-breaking) 5 A surrounding system of diverse capabilities, whether to find solutions (e.g.. diversity of firms and research institutes) or to implement them (e.g., military services) Formal and informal techniques 1 Build developmental coalitions rather than relying on (or waiting for) political will; Bring together previously unconnected capabilities, using political capital as a carrot, rather than spending that capital in command-and-control and close moni- toring 2 Distinguish “programs” and “approaches”: The first are high-level outcomes (e.g., invest in Google, build a car industry), the second are policies and investments that may or may not advance toward that outcome (e.g., add a manager to a start-up team, subsidize Hyundai). 3 For programs,bias towards consensus and caution, to cut down on broad failures; for approaches, bias toward autonomy, to cut down on missed opportunities. Use a tool like the Heilmeier Catechism for both decisions. 4 Keep the agency small: Do not grow past about 100, and stay smaller if possible. Maintain a bias toward operational expe- rience (bureaucratic or industrial) over theory, although not dogmatically 5 Over-invest in obtaining, sharing and using information, formal and informal: Use frequent (weekly) meetings focused on problems not process, and “observer rights” or the equivalent. 6 Use quantitative goals as a tool of people management, rather than a substitute for it: Use simple, clear but difficult goals to orient action, bring information to the surface, trigger problem-solving, and adjust opinions, primarily about people and capabilities (who more than what). 7 Faced with a potential failure, in order: Bring more information to the surface; add new capabilities; adjust the team or coalition; and find a new, equally ambitious goal. Only then exit. Use tools such as the questions: Can it still be a home run? If yes, what can we fix? Whom can we bring? 8 Keep quiet at first, building a record of success that can cushion the political costs of exit before exposing the agency to attack. 9 Enforce entry to prevent the risk of capture: Ensure that periods when a single entrant has monopolized a capability or industry are short, supporting the entry of credible competitors. This note is sourced from “Flexible Implementation: Techniques in Venture Capital, Defense Research, and Industrial Policy” (submitted to Policy Studies). Key References: Cazes, B., & Mioche, P. (1990). Modernisation ou décadence : études, témoignages et documents sur la planification française. Université de Provence Criscuolo, A. Palmade, V. (2008). Reform Teams: How the Most Successful Reformers Organized Themselves. Public Policy for the Private Sector Note no. 318. The World Bank. Hirschman, A. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Harvard University Press. Kim, P. G. (2011). The Park Chung Hee Era: The Transformation of South Korea. B. K. Kim, & E. F. Vogel (Eds.). Harvard University Press. Rodrik, D. (2008). The New Development Economics: We Shall Experiment, but How Shall We Learn? HKS Working Paper No. RWP08-055. Sah, R. K. Stiglitz, J. (1986). The Architecture of Economic Systems: Hierarchies and Polyarchies. American Economic Review, Vol. 76, No. 4. Wilson, J. (1989). Bureaucracy. NY: Basic Books. Disclaimer CI Notes are produced by the Competitive Industries Practice at the World Bank Group. The series aims to share lessons learned in interventions foused on boosting industry competitiveness. CI Notes are funded by the: The findings, interpretations, and conclusions expressed in this paper are those of the author(s) and do not necessarily reflect the views of the CIIP Partners, the World Bank Group, the Executive Directors of The World Bank or the governments they represent. Comptetitive Industries Note | October 2013 • Issue 3 4