Organizational Learning and Knowledge

1
Organizational Learning and Knowledge

• Charles Weber
• Innovation Management
• PSU-ETM EMGT 510/610 INNO

2
Questions Regarding Organizational Learning

• We know that individuals learn, but do
  organizations learn?
• How does an organization learn?
• How do we measure organizational learning?

3
The Learning CurveArgote Epple (1990)

• The unit cost of production typically decreases
  at a decreasing rate as organizations produce
  more of a product.
• Observed in
• Airplane Construction (Wright, 1936 Alchian,
  1963)
• Manufacturing machine tools (Hirsch, 1952)
• Refining petroleum products (Hirschmann 1964)
• The production of ships (Rapping 1965)
• Construction of power plants (Zimmerman 1982
  Joskow and Rose 1985),
• Increased production skills
• Learning by doing (Arrow, 1962)

4
Quantifying the Learning Rate

• Assumption You improve performance by investing
  in learning.
• Learning occurs according to dimensions of
  merit.
• A performance metric is a function of a proxy for
  the learning investment (which ostensibly
  represents knowledge).
• For example, Cn a n-b
• Cost is a function of cumulative output
• Cn denotes the cost of the nth unit
• a, a constant, is the cost of the first unit
  produced a gt 0
• b, a constant, represents the learning
  elasticity 0 lt b lt 1

5
The Progress Ratio

• p 2-b
• where p denotes the fraction of the unit cost
  that results from a doubling in cumulative
  output.
• A progress ratio of 0.8 or 80 implies that the
  production costs decrease by 20 by the time the
  cumulative output doubles.
• Thus if the 10th jet airliner emerging from an
  assembly line costs 10 million to build, then
  the 20th airliner will cost 8 million.

6
Variability in Learning Rates(Argote Epple,
1990)

• Generally results from
• Organizational Forgetting
• Employee Turnover
• Knowledge Transfer
• Scale Economies

From Dutton Thomas (1984)
7
Organizational Forgetting

• Unit costs may rise because ..
• of labor interruptions caused by strikes.
  (Hirsch, 1952 Baloff, 1970)
• knowledge acquired through learning by doing in
  production depreciates rapidly. (Argote, Beckman
  Epple, 1990)
• Technical information has a half life in
  entrepreneurs, which can significantly impact the
  technical base of new enterprises. (Roberts 1991,
  pp. 115-118)

8
Employee Turnover in Small Companies

• Roberts (1991, ch. 4) concludes that.
• Sixty-three out of 121 spin-off firms, which to
  varying degrees depended upon technology
  developed at the entrepreneurs previous
  employer, were founded within the first year of
  departure from the previous employer.
• The principle dissipative influence on
  technology transfer is a delay between
  terminating employment at a source organization
  and establishing a new enterprise. (Roberts,
  1991, p. 122)
• The decaying effect on the technological basis
  of the company is strong and nearly immediate,
  essentially full dissipation of transferability
  occurring within four years of departure.
  (Roberts, 1991, p. 122)

9
Knowledge Transfer

• Intra-firm knowledge transfer is not costless.
• Teece (1977)
• studied 26 international technology transfer
  projects.
• Transfer costs varied widely ranging from 2
  percent to 59 percent of the total project costs
• Average 19 percent of the total project costs
• In addition, the receiving entities of the
  technology transfer frequently exhibit
  deficiencies in quality or productivity with
  respect to the source (Teece, 1976 Mansfield et
  al., 1983),
• Some even fail to achieve profitability
  (Galbraith 1990).
• In some instances transfer of new knowledge
  dislocates the performance of the receiving
  entity (Hatch Mowery, 1998).
• Intra-firm knowledge transfer is not effortless.
• Many unforeseen events can occur. (Szulanski,
  1996, 2000)

10
Learning versus Scale

• Economies of Scale reduce unit cost by making
  more units.
• Learning by doing Reduce unit cost by
  increasing production skills.
• Hirschs (1952) study on machine tools
  manufacturing
• Small lots faster learning
• Large lots more scale
• Learning and economies of scale can be separate
  phenomena.

AC1 and AC2 represent two average cost curves at
different levels of knowledge. (from Pindyck
Rubinfeld, 1998)
11
What is going on here?
From Abernathy Wayne (1973)
12
Learning versus Innovation

• Focusing on one product mass production
  accelerates the learning rate.
• Innovation distracts from learning, thereby
  decelerating the learning rate.
• This is a part of the Productivity Dilemma
  (Abernathy, 1978).

13
Limitations of the Learning Curve

• The classical learning curve does not explain
• Learning to Learn
• Has been studied in the manufacture of machine
  tools (Hirsch, 1956) and Olympic sports (Fellner,
  1969)
• Cumulative output is a good proxy for the
  learning investment in processes that consist of
  simple tasks
• Accumulated time is a good proxy for the learning
  investment in processes that consist of complex
  tasks, which require learning how to learn.
• Indirect Learning,
• Transforms the goals of the process by explicit
  managerial and engineering action (Argyris
  Schon, 1978 Adler Clark, 1991).
• Learning before doing (Pisano, 1996) ,
• Can occur before product introduction,
• Is common in industries with large existing
  knowledge bases such as pharmaceuticals and
  semiconductors,
• Whereas learning by doing dominates industries
  with small existing knowledge base like biotech.

14
Lessons from the Lockheed L-1011 TriStar Program

• Reinhart (1973)
• used Net Present Value (NPV) as a performance
  metric and time as a proxy for the learning
  investment?
• t
• NPV(t,?) ? R(t) - C(t) e-? t dt
• 0
• t is the time since venture inception t
  investment horizon.
• ? represents the continuously compounded
  discount rate (Opportunity cost of capital).
• Conclusions
• The discount rate matters. It doomed the
  TriStar from the beginning.
• One needs to learn how to increase the revenue
  rate R(t)
• as well as to cut cost outlay rate C(t).

15
Costs, Revenues and Profitsfor Manufacturing
Microprocessors
Cost of Ownership models (e.g. Carnes Su, 1991
Martinez et al., 1992 Secrest Burggraaf, 1993
Doering, 1994 Dance Jimenez, 1994) determine
operating costs (cash outlays, C(t)).
16
Time LeverageAccelerated Learning(Weber, 2002,
2003)

• Accelerating the learning rate has the most
  pronounced effect on profitability.
• One minute delay on the critical path can amount
  to as much as 5000.
• See semiconductor process lifecycle for
  microprocessor manufacturing below.
• Yield driven (Bohn Terwiesch, 1999)

Moores Law (Moore, 1975) determines schedule.
17
Learning and Problem Solving

• Learning in most industrial settings results
  from problem-solving activity that is triggered
  by gaps between desired and actual levels of
  performance. (Newell Simon, 1972 Iansiti
  Clark, 1994 Pisano, 1996)
• Learning takes place through iterative cycles of
  search and selection, where each cycle narrows
  the absolute gap between actual and desired
  performance. (Frischmuth Allen, 1969 Nelson
  Winter, 1977 Nelson, 1982 Pisano, 1996 Thomke,
  1998)
• Problem solving consists of trial and error (or
  more precisely trial, failure, learning, revision
  and re-trial) directed by some amount of insight
  as to the direction in which a solution might
  lie. (Baron, 1988, Ch. 4 von Hippel, 1994)
• This insight tends to be more developed in
  experienced problem solvers, who typically
  outperform novices with respect to
  problem-solving speed by relating the task at
  hand to similar problems from their experience
  base. (Larkin et al., 1980 Chi et al., 1981)

18
Ill-structured Problems

• Unfortunately, most problems encountered in
  product and process development are ill
  structured (Reitman, 1965 Simon, 1973 Pople,
  1982, Ch. 5).
• They contain ambiguity as well as uncertainty
  (Schrader, 1993)
• They do not possess a known 'solution space' (a
  domain in which the solution is known to lie).
• They may also involve unknown or uncertain
  alternative solution pathways,
• They may exhibit no obvious connections between
  means and ends.
• Ill-structured problems are solved by generating
  several alternative solutions, which may or may
  not be the best possible solutions -- one has no
  way of knowing (von Hippel, 1994).
• These alternatives are then tested against a
  whole array of requirements and constraints
  (Marples, 1961 Simon, 1981, p. 149), which is a
  time-consuming process.

19
Learning and Knowledge

• Learning is the acquisition of knowledge.
• But what is knowledge?
• Are there different types of knowledge?
• How do data and information differ from
  knowledge?

20
The Ancient Greeks on Knowledge(Ihde, 1993
Spender, 1996)

• Noesis absolute (Platonic) truth
• Dianoia mathematically proven
• Pistis perceptions and beliefs
• Eikasia images of concrete objects
• Techne how to get practical things done
• Phronesis understanding social activity and
  politics
• Metis shrewdness or cunning

21
Some Western Thinkers on Knowledge (from Bohn,
1994)

• Knowledge is power. Francis Bacon
• When you can measure what you are speaking
  about, and express it in numbers, you know
  something about it but when you cannot measure
  it when you cannot express it in numbers your
  knowledge is of a meager and unsatisfactory kind
  it may be the beginning of knowledge, but you
  have scarcely, in your thoughts, advanced to the
  stage of science. Lord Kelvin (1890s)

22
Pragmatic Knowledge(James, 1950 Spender, 1996)

• Know how
• The capacity to act
• I know how to.
• savoir (French)?
• wissen (German)?
• Knowing about

• Know what
• Certain information
• I know that..
• connaître (French)?
• kennen (German)?
• Knowledge of acquaintance

Science is the process of purification, which
renders knowledge of acquaintance into
knowledge about. (J. C. Spender, 1996)
23
Knowledge Explicit or ImplicitWe know much
more than we can explain.(Polanyi, 1962, 1966)

• Explicit
• Is easy to encode, document or articulate.
• Similar to knowledge about
• Derived from positivist science.
• By formulating logical hypotheses and performing
  repeatable tests.
• Example writing a book generates explicit
  knowledge.

• Implicit (Tacit)
• Is difficult to encode, document or articulate.
• Associated with experience
• Comes from deep immersion in the phenomena to be
  explained.
• Involves intuition
• Examples breathing, riding a bicycle.

24
Tacit Knowledge and Competitive Advantage

• Kogut and Zanders (1992) paradox
• Explicit knowledge is easy to imitate.
• Tacit knowledge is difficult to imitate.
• Explicit knowledge is easy to transfer.
• Tacit knowledge is difficult to transfer.
• I need to keep my technology a secret as long as
  possible to prevent imitation, but I need to
  transfer and ramp up to production in a really
  short time. What do I do?
• Can you give an example of Kogut and Zanders
  paradox from personal experience?

25
Sticky Information (von Hippel, 1994)

• Information that is costly to acquire, transfer
  or use.
• Can be embodied in mind (tacit) or equipment.
• Determines the locus of problem solving.
• Interferes with specialization (Weber, 2002)
• Task partitioning between users and suppliers of
  a technology.
• Causes iteration between locations during problem
  solving.
• Information can be unstuck by using a toolkit
  The logic of ASIC. (von Hippel, 1998, 1999).
• Suppliers provide design tools and rules.
• Users can design without understanding supplier
  process.
• Supplier can produce many products with one
  process.

26
Knowledge Creation Theories(Nonaka, 1994 Nonaka
Takeuchi, 1995)

• Question How many of these mechanisms truly
  create knowledge as opposed to transferring or
  converting it?

27
Knowledge Creation under Time Pressure(Weber,
2002)

• Time pressure increases the stickiness of
  information.
• No opportunity for socialization.
• Tacit knowledge cannot be transferred rapidly
  enough.
• Tacit knowledge concentrates in the minds of a
  few experts.
• Both problem-specific and context-specific
  knowledge.
• Demand for these experts increases.
• New experts cannot be generated rapidly enough.
• Dynamic could cause stagnation in the
  semiconductor industry.

28
The Knowledge-Based View of the Firm (Kogut,
Grant, Spender, Liebeskind, Nonaka, von Krogh)

• Firms are stockpiles of knowledge.
• This knowledge gives them capabilities.
• These capabilities can be observed by actions.
• Capabilities are the source of competitive
  advantage.
• Key debates
• How do firms convert knowledge into competitive
  advantage?
• Do different firms have different capacities to
  absorb knowledge? (Cohen Levinthal, 1990)

29
Data, Information Knowledge

• Context
• These pictures represent composite images of
  semiconductor wafers.
• Events constitute defects.
• Certainty
• The green defects come from a spray nozzle on a
  rinser.
• The red defects are mechanical scratches that
  come from a lithography tool.
• Capacity to Act
• We have people that know how to fix these
  problems.

30
Bohns (Engineering-Oriented) Definitions Of
Data, Information and KnowledgeFrom R. Bohn,
"Measuring and Managing Technological Knowledge,
Sloan Management Review, Fall 1994, pp. 61-62.

• Data are what comes directly from sensor
  reporting on the measured level of some
  variable.
• Information is data that have been organized or
  given structure.
• Information tells the current or past status of
  of some part of the production system.
• Knowledge goes further it allows the making of
  predictions, causal associations, or prescriptive
  decisions about what to do.

From R. Glazer, Marketing in an
Information-intensive environment Strategic
implications of knowledge as an asset, Journal
of Marketing 55 (1991), pp. 1-19.
31
General Definitions of Data, Information and
Knowledge

• Data recorded events.
• Information
• Data that exhibit a discernable pattern
• Information can be quantified. (Shannon
  Weaver, 1949)
• Knowledge
• Certain information (Shannon Weaver, 1949)
• Justified true belief (neo-Kantian)
• Justification comes from context.
• The capacity to act (operational, Stehr, 1992)

32
ApplicationBohns Eight Stages of Knowledge
From R. Bohn, "Measuring and Managing
Technological Knowledge, Sloan Management
Review, Fall 1994, p. 63.
33
Discussion Automating a Plant

• Your company wants to automate an industrial
  process. What should it take into consideration?
• What is the purpose of automation?
• What about organizational forgetting and employee
  turnover?
• What are the effects of problem structure?
• Time pressure? How fast can data be converted
  into information and knowledge?
• Can tacit knowledge be automated?
• Are the preconditions for automation similar to
  those for technology transfer?
• Does automation interfere with innovation?
• Does automation interfere with new product
  development?

34
Summary The Effects of Knowledge Stages
From R. Bohn, "Measuring and Managing
Technological Knowledge, Sloan Management
Review, Fall 1994, p. 68.