1 Introduction
There is a substantial body of literature dealing with the analysis and modelling of firm
mortality and survival (for literature reviews: see e.g. Aldrich and Marsden, 1988, or
Hannan et al., 1998). An important factor in these models is the age dependency of the
mortality rate. This originates from the hypothesis, first stated in organizational
research, about the liability of age (Stinchcombe, 1965): new organizations (firms) fail
at a higher rate than old ones. This may be due to various underlying mechanisms. One
asserts that young firms have not yet built up substantial internal strength, trust, tacit
knowledge, networks with other firms and institutions, and are therefore more
vulnerable for failure. As firms age, they learn, build up strength and trust, and know
how to deal with difficulties. Another explanation is selection: as the cohort ages, the
frail organizations die out, and the surviving population therefore become on average
healthier). Although this age-dependence is a well-established stylised fact, various
authors have proposed other, age-related ‘liabilities’. The liability of adolescence states
that the mortality rate first increases, and next, after the period of adolescence, decreases
over the rest of the life span (Brüderl and Schüssler, 1990). The liability of
obsolescence states that there is a positive relationship between age and mortality. In
this reasoning, organizational inertia leads to a position which is increasingly ‘out of
sink’ with the environment, and hence to a larger vulnerability to failure (Barron et al.,
1994). The liability of senescence asserts that organizations itself change as they age,
becoming less and less flexible, more bureaucratic and less efficient, with a similar
effect of increasing mortality with higher age (Carroll and Hannan, 2000). All these
liability hypotheses are propositions about the form of the age curve of mortality.
Empirical research has given overwhelming evidence for decreasing mortality with
increasing age, thus -at least for most real-world situations- ruling out the obsolescence
and senescence hypotheses.
The emphasis on age gives these types of research a strong demographic flavour. Yet,
from a demographic point of view, these analyses are incomplete. In demography, time
is the fundamental dimension. In fact, it is so fundamental, that it has been divided into
a number of dimensions itself. The most common decomposition of time is in the
dimensions age, period and cohort. Age is the amount of time elapsed since birth;
Period is chronological or calendar time, and cohort is the chronological time during the
formative period. An exclusive focus on age in the description of mortality gradients of
firms may disregard important influences that are linked to specific historical events
(e.g. economic recessions, wars) or that are related to specific cohort differentials. The
demographic toolbox has a solution to this problem, which is called the age-period-
cohort (APC) model. This model decomposes variations across age groups over time,
into these three dimensions.
The dimension of calendar time as a source of variation in firm survival has also
received some attention, most often in the form of analyses of the relation ship between
the economic business cycle and survival. Indeed, there is a positive relationship
between economic growth and firm survival (Siegfried and Evans, 1994, Caves, 1998).
However, the relationship between calendar time and survival, without taking into
account the possible impact of the age distribution of firms, may give biased results. For
instance, in times of economic growth, entry will be high, and due to the short-lived
nature of most firms, exit will be high as well. In order to solve this problem, age and
time should be taken into account simultaneously. But even then, the cohort dimension
may be important as well.