Path dependence

Path-dependence explains how the set of decisions one faces for any given circumstance is limited by the decisions one has made in the past, even though past circumstances may no longer be relevant. [Definition from [http://poopthebook.com/blog/?p=20 "Our Love Of Sewers: A Lesson in Path Dependence"] , Dave Praeger, 15 June 2008.]

The phrase is regularly used to mean one of two things (Pierson 2004):
*Some authors use path-dependence to mean simply "history matters" - a broad concept;
*Others use it to mean that institutions are self reinforcing - a narrow concept.It is the narrow concept that has the most explanatory force and of which the discussions below are examples. The claim "history matters" is trivially true and reduces simply to "everything has causes".

Illustration

Consider as an example the technological development of videocassette recorders (VCRs) for home use. It is argued that management errors and minor design choices by Sony, was one of the reasons why its Betamax format being defeated in market competition by VHS in the 1980s. Two mechanisms can explain why the small but early lead gained by VHS became larger over time. The first is the bandwagon effect of VCR manufacturers in favor of the VHS format in the U.S. and Europe, who switched because they expected VHS to win the "standards battle". The second was a network effect: videocassette rental stores observed that more people had VHS players and stocked up on VHS tapes; this in turn led other people to buy VHS players, and so on until there was complete vendor lock-in to VHS. An alternative explanation, of course, is that VHS was better adapted to market demands (in particular to the demand for longer cassettes for recording sports games) and that path dependence had little or nothing to do with its success. There is also some support for this latter claim.

Positive feedback mechanisms like bandwagon and network effects are at the origin of path-dependence. They lead to a reinforcing pattern, in which industries 'tip' towards one or another product design. Uncoordinated standardisation can be observed in many other situations.

Examples from economics, history, software, and biology are presented below.

Economics

Path dependency theory was originally developed by economists to explain technology adoption processes and industry evolution. The theoretical ideas have had a strong influence on evolutionary economics (e.g., Nelson & Winter 1982).

There are many models and empirical cases where economic processes do not progress steadily toward some pre-determined and unique equilibrium, so that the nature of any equilibrium achieved depends partly on the process of getting there. The outcome of a path dependent process will often not converge towards a unique equilibrium but instead reach one of several equilibria (sometimes known as absorbing states).

This dynamic vision of economic evolution is very different from the neo-classical economics tradition, which in its simplest form assumed that only a single outcome could possibly be reached, regardless of initial conditions or transitory events. With path dependence, both the starting point and 'accidental' events (noise) can have significant effects on the ultimate outcome. In each of the following examples it is possible to identify some random events that disrupted the ongoing course, with irreversible consequences:

* In the 1980s, the U.S. dollar exchange rate appreciated, lowering the world price of tradable goods below the cost of production in many (previously successful) U.S. manufactures. Some of the factories which closed as a result could now be run at a (cash-flow) profit, because the dollar has declined. However, re-opening them is too expensive. This is an example of hysteresis, switching barriers, and irreversibility.

* In economic development, it is said (initially by Paul David in 1985) that a "standard" which is first-to-market can become entrenched (like the QWERTY layout in typewriters still used in computer keyboards). He called this "path dependence", and argued that inferior standards can persist simply because of the legacy they have built up. The case against QWERTY has been criticized (e.g. by "The Fable of the Keys" [Liebowitz, S.J. and Stephen E. Margolis (1990), [http://www.utdallas.edu/~liebowit/keys1.html "The Fable of the Keys"] , "Journal of Law & Economics" vol. XXXIII (April 1990)] ), but standards are clearly very important in modern economies, and the significance of path dependence in determining how they form is the subject of economic debate.

* Economists since Adam Smith have noted that businesses of a certain type tend to congregate geographically, attracting workers with skills in that business, which draw in more businesses looking for employees with experience. There may not have been any particular reason to prefer one place to another before the industry developed, but as it has become concentrated in one place any new entrants elsewhere are at a disadvantage, and will tend to move into the hub if possible, further increasing its relative efficiency. The system which causes this is a network effect, ideas of which lead to the New trade theory and Krugman's "New Economic Geography" are based partly on this story.

* If the economy follows adaptive expectations, future inflation is partly determined by past experience with inflation, since experience determines expected inflation and this is a major determinant of realized inflation.

* A transitory high rate of unemployment during a recession can lead to a permanently higher unemployment rate because of the skills loss (or skill obsolescence) by the unemployed along with a deterioration of work attitudes. In other words, cyclical unemployment may generate structural unemployment. The negative effects get reinforced by potential employers' negative view of the capacities of job-seekers who have been out of a job for a long time. This structural hysteresis model of the labour market differs from the prediction of a "natural" unemployment rate or NAIRU, around which 'cyclical' unemployment is said to move randomly. Since structural unemployment is endogenous, the NAIRU is also endogenous (see the article by Hargreaves Heap cited below).

Liebowitz and Margolis distinguish between different types of path dependence. Some types of path dependence do not imply inefficiencies and, while they may be interesting to study for other reasons, do not challenge the policy implications of neoclassical economics. Only what they call "third degree" path dependence - for example, a situation where society would be better off if everybody switched standards simultaneously, but they do not do so because there is no central authority to force them to, and they cannot all coordinate - involves such a challenge. They argue that such situations can be expected to be rare for theoretical reasons and that this prediction is borne out by what they consider the unconvincing examples typically discussed in this context (mainly VHS vs. Beta and QWERTY vs. Dvorak).

In technical terms, a path-dependence (stochastic system) can be defined as "one possessing an asymptotic distribution that evolves as a consequence (function of) the process's own history". This is also known as a "non-ergodic stochastic process". Confusingly, the use of "path dependent" to describe labour market hysteresis has the "opposite" sense to the term's meaning in the adaptive expectations model of inflation. In labour market economics, some "path dependent" models have unemployment following a driftless random walk, based solely on its previous level (a Markov process).

History and the social sciences

The history of humanity is almost by definition path-dependent. Accidental events such as the death at an early age of major historical figures like Napoleon or Hitler would likely have altered the political geography of Europe and even the languages spoken in different countries today.

Recent methodological work in comparative politics and sociology has adapted the concept of path dependence into analyses of political and social phenomenon. Path dependence has primarily been used in comparative-historical analyses to analyze the development and persistence of institutions, whether they be social, political, or cultural. There are arguably two discernible types of path-dependent processes:

* One is the "critical juncture" framework, most notably utilized by Ruth and David Collier in political science. In the critical juncture framework, antecedent conditions define and delimit agency during a critical juncture in which actors make contingent choices that set a specific trajectory of institutional development and consolidation that is difficult to reverse. This is akin to the concepts of vendor lock-in or positive feedback derived from path dependence in economics.

The critical juncture framework has been used to explain the development and persistence of welfare states, labor incorporation in Latin America, and the variations in economic development between countries, among other things.

An influential attempt to give some formal rigor to thinking about path dependence in political science is notably that of Paul Pierson. Pierson draws in part on ideas from economics (see above). His efforts in this regard have been questioned by Herman Schwartz, who argues that forces analogous to those identified in the economic literature are not pervasive in the political realm, where larger forces and the strategic exercise of power give rise to, maintain, and transform institutions.

In a related vein, scholars such as Kathleen Thelen caution that the historical determinism in path-dependent frameworks ignore the constant renegotiation of institutional configurations. She suggests that institutions undergo moments of institutional evolution wherein key actors renegotiate the configuration and purpose of institutions.

* The other path-dependent process deals with "reactive sequences" where a primary event sets off a temporally-linked and causally-tight chain of events that is nearly uninterruptible. These reactive sequences have been used to link the death of Martin Luther King, Jr. with welfare expansion and the industrial revolution in England with the development of the steam engine.

Other examples

*A general type of path dependence is a typological vestige.
**In typography, for example, some customs persist, although the reason for their existence no longer applies; think for example of the placement of the . inside a quotation in U.S. spelling.
*Evolution is considered by some to be path-dependent: random mutations occurring in the past have had long-term effects on current life forms, some of which may no longer be adaptive to current conditions. For instance, there is a controversy about whether the panda's thumb is a leftover trait or not.
*In the computer and software markets, legacy systems indicate path dependence: customers' needs in the present market often include the ability to read data or run programs from past generations of products. Thus, for instance, a customer may need not merely the best available word processor but rather the best available word processor that can read Microsoft Word files. Such limitations in compatibility contribute to lock-in, and more subtly, to design compromises for independently developed products if they attempt to be compatible. It is not clear, however, that there is any inefficiency involved in the costs of remaining compatible with past decisions.

Notes

References

* Arrow, Kenneth J. (1963), 2nd ed. "Social Choice and Individual Values". Yale University Press, New Haven, pp. 119-120 (constitutional transitivity as alternative to path dependence on the status quo).
* Arthur, W. Brian (1994), "Increasing Returns and Path Dependence in the Economy", Ann Arbor, Michigan: University of Michigan Press.
* Boas, Taylor C. (2007), [http://socs.berkeley.edu/~tboas/pathdependence.pdf "Conceptualizing Continuity and Change: The Composite-Standard Model of Path Dependence"] , Journal of Theoretical Politics 19(1): 33-54.
* David, Paul A. (2000), [http://econwpa.wustl.edu/eps/eh/papers/0502/0502003.pdf "Path dependence, its critics and the quest for ‘historical economics’"] , in P. Garrouste and S. Ioannides (eds), "Evolution and Path Dependence in Economic Ideas: Past and Present", Edward Elgar Publishing, Cheltenham, England.
* Hargreaves Heap, Shawn (1980), "Choosing the Wrong 'Natural' Rate: Accelerating Inflation or Decelerating Employment and Growth?" "Economic Journal" 90(359) (Sept): 611-20 (ISSN 0013-0133)
* Mahoney, James (2000), “Path Dependence in Historical Sociology,” Theory and Society 29:4, pp. 507-548.
* Stephen E. Margolis and S.J. Liebowitz (2000), [http://www.utdallas.edu/~liebowit/palgrave/palpd.html "Path Dependence, Lock-In, and History"]
* Nelson, R. & S. Winter (1982), "An evolutionary theory of economic change", Harvard University Press.
* Pierson, Paul (2000). "Increasing Returns, Path Dependence, and the Study of Politics." American Political Science Review, June.
* _____ (2004), "Politics in Time: Politics in Time: History, Institutions, and Social Analysis", Princeton University Press.
* Puffert, Douglas J. (1999), [http://eh.net/encyclopedia/article/puffert.path.dependence "Path Dependence in Economic History"] (based on the entry "Pfadabhängigkeit in der Wirtschaftsgeschichte", in the "Handbuch zur evolutorischen Ökonomik")
* _____ (2001), [http://www.vwl.uni-muenchen.de/ls_komlos/spatial1.pdf "Path Dependence in Spatial Networks: The Standardization of Railway Track Gauge"]
* Schwartz, Herman. [http://www.people.virginia.edu/~hms2f/Path.pdf "Down the Wrong Path: Path Dependence, Increasing Returns, and Historical Institutionalism."] , undated mimeo
* Shalizi, Cosma (2001), [http://cscs.umich.edu/~crshalizi/notebooks/qwerty.html "QWERTY, Lock-in, and Path Dependence"] , unpublished website, with extensive references