Professional Documents
Culture Documents
When Does Prior Experience Pay (2014) Perkins, S, L
When Does Prior Experience Pay (2014) Perkins, S, L
59 (1)145–181
When Does Prior Ó The Author(s) 2014
Reprints and permissions:
Experience Pay? sagepub.com/
journalsPermissions.nav
Susan E. Perkins1
Abstract
This study reexamines organizational learning theories to reconcile the condi-
tions under which prior internationalization experience leads to performance
gains for multinational corporations (MNCs) with varying host-country institu-
tional experiences in different regulatory environments. Using field studies on
telecommunications regulation, executive interviews conducted in Brazil,
Spain, Portugal, Canada, and the U.S., and foreign direct investment data for 96
subunit operations investing in the Brazilian telecommunications industry from
1997 to 2004, I develop an experiential-learning theoretical framework to
explain the mechanisms driving MNCs’ performance in subsequent host-
country institutional environments given the prior experience they acquired in
80 heterogeneous regulatory environments. I predict and find that MNCs with
highly similar institutional experience compared with the target country’s insti-
tutional environment will succeed. Empirical evidence suggests that similarity,
breadth, and depth of prior regulatory experience significantly prolong survival.
In contrast, firms with institutional experience unrelated to the target country’s
regulatory environment experience learning penalties and are six times more
likely to fail. These findings suggest that variations in learning contexts affect
organizations’ learning curves.
1
Kellogg School of Management, Northwestern University
146 Administrative Science Quarterly 59 (2014)
1997). When the learning environment is stable, these findings parallel more
general experiential learning theories that predict performance benefits (e.g.,
productivity, quality, survival) as the overall level of experience increases (Levitt
and March, 1988; for a review, see Argote, 1999). Several empirical studies in
the organizational learning literature demonstrate similar learning effects of
prior experience mainly by replication of routines in the same contextual envi-
ronment (Dutton, Thomas, and Butler, 1984; Darr, Argote, and Epple, 1995;
Epple, Argote, and Murphy, 1996).
Theoretical predictions are more questionable, however, when the environ-
mental context of learning changes. Institutional environments, generally char-
acterized by idiosyncratic laws, regulations, political hazards, and cultural
norms, are known to affect strategic decision making (e.g., Kogut and Singh,
1988; Henisz and Delios, 2004; Zhao, 2006; Siegel and Larson, 2008) and per-
formance outcomes (Barkema, Bell, and Pennings, 1996; Shaver, Mitchell, and
Yeung, 1997; Henisz and Delios, 2004; Berry and Sakakibara, 2008). In the glo-
bal context, where keen distinctions define heterogeneous country-level institu-
tional environments (see Henisz, 2004; and Guillén and Suarez, 2005, for
reviews), it is unclear how prior experiences acquired in other countries contrib-
ute to multinational corporations’ (MNCs’) performance in new host countries.
Despite the growing body of knowledge about organizational learning in the glo-
bal context, this puzzle remains unsolved. What firms learn from heteroge-
neous host-country environments, and whether all types of prior experience
have the same propensity to lead to future performance improvements,
remains unknown. This study builds on experiential learning theories in both
the organizational learning and international business literatures to explore the
boundaries of learning across heterogeneous institutional contexts and to
examine the effects on subsequent performance. I unravel the conditions in
which prior experience leads to positive performance outcomes and quantify
the learning penalties from unrelated institutional experience.
Experience is most broadly conceptualized as an organization’s historical
memory of routines (Cyert and March, 1963; Nelson and Winter, 1982; Levitt
and March, 1988) that, when retrieved and replicated, improve organizational
performance by reducing uncertainty. Organizational benefits derived from
accumulated experience are collectively referred to as organizational learning
curves (Yelle, 1979; Argote and Epple, 1990; Argote, 1999) or experience-based
learning curves (Huber, 1991: 94). Organizations’ prior experience with routi-
nized tasks in a homogenous context is associated with performance improve-
ments derived from learning-curve benefits in settings as diverse as U.S.
aircraft manufacturing (Dutton, Thomas, and Butler, 1984), shipbuilding
(Argote, Beckman, and Epple, 1990), truck assembly (Epple, Argote, and
Devadas, 1991), shift workers (Epple, Argote, and Murphy, 1996), pizza produc-
tion (Darr, Argote, and Epple, 1995), and cardiovascular surgical procedures
(Pisano, Bohmer, and Edmondson, 2001).
In the international context, foreign investment theorists have examined the
role of prior experience in organizational learning. The seminal theoretical per-
spectives established by scholars such as Johanson and Vahlne (1977) and
Kogut (1983) suggest that MNCs’ investment capacity depends on the sequen-
cing patterns and deployment of knowledge gained from previous investments.
Such experiences of internationalization allow organizations to acquire market-
specific knowledge that decreases uncertainty, which is a constraint on
Perkins 147
industry from 1997 to 2004. By using field studies, interviews with executives
and regulators in Brazil, Spain, Portugal, Canada, and the U.S., and a foreign
investment database, I develop an experiential learning framework to codify
and examine the effects of MNCs’ prior home and host-country experiences on
subsequent investment in Brazil.
Institutional Similarity
Each institutional experience can be thought of as a type of knowledge the
organization acquires across a range of institutional environments. In a host
country, the institutional environment includes multiple dimensions from which
an organization learns. For example, dimensions of formal government regula-
tion include the statutory laws of the competitive market structure, entry bar-
riers, and industry standards, as well as the regulators’ power to influence the
market or the stability of the regulatory governance structure. Akin to these are
the widely explored informal institutional dimensions of culture, including indivi-
dualism versus collectivism, masculinity versus femininity, power distance, etc.
(Hofstede, 1980). As a firm’s investment portfolio expands into new countries,
the scope of experiential knowledge types likely broadens.
Experiential learning studies in the international business and strategy litera-
tures have predicted that familiarity with relevant dimensions of the institutional
environment leads to subsequent success in similar types of environments.
Delios and Henisz (2003) found that firms with more experience in countries
with high political hazards are less sensitive to such hazards in their subsequent
foreign entries. Barkema, Bell, and Pennings (1996) used cultural distance to
150 Administrative Science Quarterly 59 (2014)
But not all types of knowledge acquired from multiunit organizations’ experi-
ences are relevant when transferred between subunits. Although recent organi-
zational learning studies have demonstrated that similarity between prior and
current experience has positive effects on performance, learning discounts also
exist. For example, through SIC code matching, Haleblian and Finkelstein
(1999) found that acquisition success is highly correlated with similar prior
acquisition experience in the same industry, but acquiring firms are more prone
to fail when new acquisition partners have dissimilar types of acquiring experi-
ence. Likewise, Ingram and Baum (1997) found that Manhattan hotel chains
with nonlocal experience were more likely to fail despite the benefits gained
from hotel chain affiliation. These findings are consistent with the international
business ‘‘liability of foreignness’’ theory (Zaheer and Mosakowski, 1997),
which suggests that failure risks increase with dissimilarity between foreign
and host countries’ institutional environments. Without relevant experience,
firms are more likely to grossly misestimate the effect of the institutional envi-
ronment on their business operations until they have better knowledge of it.
This ‘‘learning penalty’’ can accelerate the time to failure because inappropriate
knowledge from dissimilar contexts is applied. Thus I hypothesize:
Hypothesis 1b: The learning penalty associated with dissimilar institutional experi-
ence will be disproportional to the learning benefits of having similar institutional
experience.
market and found evidence that learning from early entries into core busi-
nesses enabled firms to launch future entries in more-risky noncore businesses
with less experience. In Erramalli’s (1991) investigation of the foreign invest-
ment entry behaviors of U.S. service firms, initial patterns of international
expansion were most evident in similar cultural environments. As organizations
gained experience with investing abroad, they became increasingly likely to
choose less-similar markets. These empirical studies confirmed Davidson’s
(1980) theoretical view that inexperienced MNCs are likely to select foreign
investment locations that are more similar to those of their host country than
firms with a broader range of investments. Organizational learning theories of
repetitive momentum (March, 1991; Amburgey and Miner, 1992) also predicted
that organizations are more likely to repeat routines with which they are famil-
iar. If the predictions of the extant literature prevail, a logical performance impli-
cation is that firms will reap the greatest benefits from reinvesting repeatedly
in the same host-country institutional environment. This strategy is bounded,
however, as investment opportunities diminish in a host country over time, and
there will always be variations at the country level. As MNCs internationalize,
the range in the types of experience acquired is likely to expand their knowl-
edge base. Expansion into institutional environments that vary significantly
from the home-country institutions increases the learning barriers for multina-
tional organizations but also possibly offers greater rewards.
Cyert and March (1963) anchored the learning argument with a more expan-
sive view by asserting that more learning experiences lead to more possible
combinations and a broader range of future choices. Applying this logic to an
MNC’s cumulative breadth of prior institutional experiences yields a reasonable
rationale that organizations that expand into new and uncertain markets have
the advantage of a greater knowledge pool. Firms potentially gain more from
these difficult learning environments, particularly in emerging market econo-
mies, which typically have more-volatile and weaker institutions. Further,
MNCs that possess greater variation in knowledge-retrieval sources are likely
to benefit from generating a greater number of combinative experiences to
leverage in future subunit investments. In another context, Beckman and
Haunschild (2002) showed that heterogeneity in the interorganizational network
structures of acquisition partners leads to better performance outcomes.
Pennings, Barkema, and Douma (1994) provided evidence that firms successful
in diversification have persistent success with subsequent firm expansions.
This suggests that organizations assuming greater risks in investing in varying
types of host-country institutional environments should reap benefits derived
from their breadth of knowledge.
Experience-based Capabilities
The inherent heterogeneity of country-specific institutional dimensions creates
a challenge for firms to identify and build in-depth capabilities across each type
of institutional experience over time. Kogut and Zander (1992) argued that the
creation of new knowledge stems from the recombination of existing capabil-
ities. These recombination opportunities exist for MNCs learning from their
152 Administrative Science Quarterly 59 (2014)
METHODS
Sample and Data Collection
The empirical setting I examined was the entire population of foreign subunit
investments in the Brazilian telecommunications industry from 1997 to 2004
following market privatization and liberalization. This setting is ideal for empiri-
cal investigation for several reasons. First, MNCs face considerable complexity
and heterogeneity in environmental contexts (Ghoshal and Westney, 2005) as
they internationalize. Second, the global telecommunications industry, rife with
government interventions and regulation, provides codifiable ‘‘rules of the
game’’ to measure learning constraints. Because of the tractable nature of gov-
ernment regulations and the fact that firms’ compliance is compulsory, reason-
able inferences can be made about organizational learning in this setting. Last,
Brazil’s economy, representative of many emerging market economies, is
Perkins 153
1
R$ denotes real dollars, Brazilian currency (reais). Subsidiary size is represented by annual sales
revenues that have been adjusted in real values based on the consumer price index (CPI) for tele-
coms in 1995 reais.
154 Administrative Science Quarterly 59 (2014)
entries in less than one year, and five were observations of exits that were not
failures (Headd, 2003).2 The remaining 779 firm/parent/year observations were
used for empirical examination.
I constructed this dataset by using a broad selection of data collection and
triangulation techniques, including several firsthand data sources in Brazil—
Conselho Administrativo de Defensa Economica (CADE), BNDES
(Development Bank of Brazil), ANATEL (the Brazilian telecommunications
regulatory agency), and Comissão de Valores Mobiliários (CVM), Brazil’s
securities and exchange commission. Secondary archival data sources
included ISI Emerging Markets, Espicom Business Intelligence, BuddeComm
Telecommunications Reports, and Dunn & Bradstreet Million Dollar,
Hoover’s, Gale, and Orbis databases. I used other publicly available periodi-
cals acquired through Factiva, company reports, and press releases to vali-
date each subunit’s entire event history and subsidiary locations for its
parent’s prior experience. I augmented these data with interview data from
30 telecommunications and regulatory agency senior executives, who
recounted market events.
Dependent Variable
I analyzed the time to failure event, a firm’s exit from Brazil, at the firm/parent/
year level, where firm survival in period t was (0) or failure was (1). Defining fail-
ure can be problematic in event history studies because some market entries
are not designed for longevity (e.g., organizational learning) (Nakamura, Shaver,
and Yeung, 1996), and some market exits are not the result of organizational
failure (e.g., capital gains) (Headd, 2003). Therefore I examined each exit to
determine the reason for it. I limited failures to exits driven by financial under-
performance in earnings projections, difficulty securing capital to continue the
investment, lack of understanding of local institutions, or related situations.
Three independent coders tested the reliability of the assessment of failure
cases. Interrater reliability was 100 percent in all cases.
Independent Variables
3
Statutory laws related to the regulatory competitive market structure are the most widely cited in
the literature on industry regulation (Stigler and Friedland, 1962; Demsetz, 1968; Koller, 1973;
Priest, 1993).
4
Brazil is a suitable reference point because telecom regulation is relatively close to the mean level
of telecom regulation among the 80 countries of prior experience examined in this study.
156 Administrative Science Quarterly 59 (2014)
and the target country’s dimensions are captured in the vector Yj such that
0 1
competitive market structureY
B standardsY C
B C
B entry barriersY C
Y =B
B political appointment processY
C
C ð6x1Þ:
B C
@ governance structureY A
institutional stabilityY
The regulatory distance scores between Brazil and other countries are pre-
sented in Online Appendix F.
3 piled up across each country such that dj 8j
2 dj, is
This regulatory distance,
D1
generates the vector D = 4 ::: 5 for all countries j. This vector provides com-
Dn
parative distance measures for each host country versus the regulatory envi-
ronment in Brazil. To make these measures meaningful at the firm level, I
recorded the host countries of experience prior to the beginning of this study
(t ≤ 1998) such that for each parent firm i, the global portfolio of investments is
2 3
0 0 1 p1n
61 0 0 p2n 7
P = pij = 6
4 ::: :::
7ðk × nÞ,
::: ::: 5
0 1 0 pkn
Regulatory
Competitive
Market Regulatory
Structure Entry Barriers
Regulatory Brazil
Standards
Mahalanobis
Distance
Parent Firmj
regulatory similarity between the firm’s experience and the regulatory environ-
ment of Brazil. Figure 1 provides a graphical illustration of this comparative dis-
tance measure.
I used three weighting approaches, described below, to capture the home-
country effects, host-country effects, and combined regulatory effects.
the firm acquires prior to the birth of a new organization. For multinationals,
new organization events are subunit investments in host countries. The other
sources of information include acquired knowledge from MNCs’ prior experi-
ences in institutional environments globally. In Huber’s (1991) knowledge acqui-
sition framing, organizations rely on congenital knowledge for information
retrieval to assist future decision making about which countries to invest in,
what inherent capabilities the firm will deploy relative to market needs, and
what resources will be needed in the targeted country’s institutional environ-
ment. To capture congenital knowledge, regulatory distance is weighted (wij) to
include the entire set of host countries a firm has previously entered, with all
host-country
P investments weighted equally. In this case, the weight wi,h = 0
and wi,j = 1translates into 100 percent of the experiential learning weight
j
being given to host countries and none to home-country experience.
Control Variables
Host-country experience and internationalization experience. To first
replicate the predicted theory and findings of the conventional wisdom on
MNC experience, I developed two variables. Host-country prior experience cap-
tures the number of subsidiary investments in Brazil. This count variable repli-
cates the methodology used in the existing literature on foreign investment
(Barkema, Bell, and Pennings, 1996), which suggests that a firm’s own experi-
ence in a host-country affects performance. The second variable, internationali-
zation experience, measures the number of foreign host countries each
multinational has entered prior to entering Brazil.
5
Hofstede’s cultural dimensions did not include Luxembourg, so I used the cultural measures for
Belgium as a proxy, given the proximity and historical ties. Kogut and Singh (1988: 422) defined cul-
tural distance using the computation
4
P
CDj = (Iij Iiu )2 =Vi g=4;
i =1
where Iij is the index for the ith cultural dimension and the jth country, Vi is the variance of the index
of the ith dimension, u indicates the U.S., and CDj is the cultural difference of the jth country from
the U.S.
6
The Euclidian distance, DE = ðpx qx Þ2 = jpx qx j, was used to calculate a distance for each par-
ent company’s home country in comparison to Brazil for the POLCON, legal origins, and language
variables. I used the simplest form of the Euclidean distance, given that each of these measures is
one-dimensional.
7
La Porta et al.’s (1998) study did not include legal origins in Panama and Luxembourg. Both of
these civil-law societies have been coded accordingly.
Perkins 161
Empirical Model
A log-logistic parametric duration-dependence model was specified to both
accommodate the time-varying covariates and account for the liability of adoles-
cence and life-cycle-oriented factors of these new subunit organizations (Carroll
and Huo, 1986; Hannan and Freeman, 1989; Brüderl and Schüssler, 1990). I
confirmed the expected nonmonotonic distribution (i.e., inverted-U shape)
using nonparametric estimates to determine the appropriateness of fit of a log-
logistic form. A time-to-fail event, exit, was predicted by
λρ(λt)ρ1
h(t) =
1 + (λt)ρ
at a given point in time t for every subunit/parent firm observation i. The covari-
ates were parameterized in l so that the functional form fit the expectation that
the covariates positively or negatively affected the baseline survival rate, a con-
venient feature of the accelerated failure time (AFT) models. A concern that
requires attention when using survival models is both left and right censoring.
The majority of subunit entries (more than 80 percent) occurred in the first two
years of market privatization, which alleviated concerns of left censoring. In the
periods prior to this study, the telecom industry was 100 percent state owned
and operated by Telebras. Right censoring is an obstacle in interpreting the out-
come of foreign subunits that are between states of success and failure. The
AFT model specification allows for greater emphasis on predictions of the orga-
nizations’ time to fail beyond the life of this study. It is also worth noting that
counterintuitive to proportional hazard rate models (i.e., Gompertz and Weibull),
AFT coefficient interpretation is consistent with the direction of the sign. That
means a positive and significant coefficient translates into a deceleration in
time to fail with each unit increase; negative coefficients accelerate time to
failure.
RESULTS
Tables 2 and 3 provide descriptive statistics for all variables and a correlation
matrix for variables included in the analysis. On average, MNCs’ prior host-
country regulatory experiences (mean = 9.3) are more proximate to the
Perkins 163
Variable 1 2 3 4 5 6 7 8 9 10 11 12
1. Cultural distance
2. POLCONV distance .417
3. Legal origin distance (LLSV) .843 .495
4. Language distance .853 .418 .833
5. Sales revenues—real (log) –.315 –.174 –.309 –.274
6. Market share (%) –.257 –.222 –.256 –.242 .639
7. Ownership (%) –.068 .104 .000 –.121 –.169 –.038
8. Business group affiliation –.729 –.424 –.781 –.616 .274 .184 –.068
9. Relative industry size .696 .392 .697 .595 –.294 –.158 .075 –.793
10. Bubble effect .002 –.012 –.055 –.047 .060 .012 –.082 –.071 .066
11. Geographic region .138 –.001 .152 .134 –.464 –.143 .266 –.147 .220 .216
12. Auction—FDI restricted .052 .365 .023 .012 .120 –.135 –.139 .013 –.051 –.156 –.477
13. Host-country experience –.432 –.323 –.504 –.502 .261 .142 –.093 .418 –.423 .464 –.150 –.046
14. Internationalization experience –.254 –.291 –.198 –.125 .106 .079 .040 .152 .006 .069 .059 –.119
15. Skills-based capabilities –.447 –.280 –.376 –.299 .231 .220 .200 .317 –.185 .059 –.019 –.064
experience
16. Fixed effects—SIC code 4812 –.046 .128 –.127 –.097 .138 –.071 –.064 .069 –.062 .039 –.204 .414
17. Fixed effects—SIC code 4813 –.014 –.131 .079 .023 –.040 .133 .053 –.011 –.002 –.106 .083 –.333
18. Fixed effects—SIC code 4822 .031 –.003 –.014 .035 –.110 –.055 –.133 –.010 .025 .000 –.008 –.050
19. Fixed effects—SIC code 4899 .111 .012 .100 .138 –.155 –.113 .074 –.116 .122 .144 .249 –.130
20. Regulatory distance— .784 .430 .806 .790 –.292 –.208 –.053 –.742 .875 .011 .174 .007
home country
21. Regulatory distance— .551 –.100 .501 .625 –.069 –.090 –.161 –.332 .200 –.021 .057 –.141
host country
22. Minimum distance .504 .331 .486 .522 –.208 –.176 –.147 –.418 .351 .047 .107 .019
23. Sign regulatory distance .428 .220 .671 .552 –.088 –.013 –.054 –.552 .527 –.075 .003 –.066
24. Regulatory distance (50/50) .821 .361 .830 .834 –.285 –.208 –.094 –.746 .824 .020 .178 –.030
25. Breadth –.136 .530 –.180 .091 .067 –.043 .040 .103 –.085 –.023 –.051 .276
26. Depth –.624 –.471 –.580 –.568 .262 .266 .015 .389 –.259 .078 –.080 –.118
Variable 13 14 15 16 17 18 19 20 21 22 23 24 25
have shown firm size is a significant predictor of success, I found that the par-
ent firm’s equity participation in the subunit firm is a more important factor, as
dominant equity owners (percentage ownership) survive longer. The entry
order variable, a proxy for asset bubbles, is consistently positive but insignifi-
cant. This suggests that perhaps later entrants fared better because they
entered after the significant industry overinvestments in acquiring licenses
were made and benefited from industry consolidation.
Model 3 measures the effects of prior foreign investment experience, opera-
tionalized as the number of host-country investments in Brazil, and internationa-
lization experience, the number of prior countries of investment. These
variables reveal some support for the argument that prior experience does pay,
as firms increase their market-specific investment commitment. Model 3
empirically replicates Shaver, Mitchell, and Yeung’s (1997) finding that foreign
entrants into U.S. manufacturing increase the likelihood of survival as the
166 Administrative Science Quarterly 59 (2014)
benefits from firms with skills-based experience. Repeat investments using the
same telecommunications technology skills (4-digit SIC code) significantly (p =
.08) decelerate time to fail. This relationship is consistently positive across all
models and becomes increasingly more significant when experience-based
measures are subsequently included. Fixed effects are also included in models
4 to 12 to account for the telecommunications technology used in each subunit
investment.
fail. The home country, regulatory distance, and sign of distance were the
most significant, at p < .001. The magnitude of the sign of distance coeffi-
cient (–0.4) is more than tenfold that of the other regulatory distance mea-
sures. This suggests that the most difficult of these learning hurdles is
learning to adapt from a highly regulated environment to a much less regu-
lated environment.
Figure 2 provides strong support for hypothesis 1b. To investigate differ-
ences in failure rates for similarity versus dissimilarity, I stratified the regulatory
distance data categorically based on the mean regulatory distance from Brazil,
holding all other variables constant at their means. Deviation below the mean
was coded (1) for similar experience and (0) for deviation above the mean for
dissimilar experience. The stratified hazard rates reveal intriguing magnitude dif-
ferences. Firms with dissimilar experience are six times more likely to fail (0.6
hazard rate) than firms with similar experience (0.1 hazard rate). Figure 2 sug-
gests that not only does similar experience pay, but the learning penalties from
dissimilar experience have disproportionally greater negative impacts on
performance.
Model 10 in table 4c provides support for hypothesis 2. Breadth of experi-
ence has a positive and significant effect on firm survival. I also conducted fur-
ther hazard-rate stratification analysis to examine whether there are conditions
in which breadth of experience can hinder performance. Similar to the analysis
above, I stratified the breadth variable to create two categories of high (1) and
low (0) breadth split on the mean and further stratified the high/low breadth
observations by the stratified distance measure above, creating four categories
in total. Results, in figure 3, reveal that firms that acquired breadth and had
experience similar to Brazil had the lowest failure rate. This suggests that firms
experience learning penalties when the breadth of experience is not relevant to
Brazil. The implication is that firms are better off having similar experiences and
limited breadth. When irrelevant institutional knowledge is introduced into dis-
similar environments, breadth penalizes performance.
Model 11 confirms hypothesis 3. Depth of prior experience in regulatory
environments matching Brazil across the different dimensions has the most
positive and significant effects on the subunit’s survival of the three hypothe-
sized institutional experience effects. The magnitude of the coefficient is six
times more powerful than breadth in decelerating time to fail. These results are
supportive of the argument that learning-curve benefits can also be achieved
from experience-based capabilities that are derived from recombining learned
routines across institutional environments in highly applicable target host
countries. The full model 12, including all control variables, conventional prior
experience variables (number of prior investments in Brazil and number of
host countries of investment), skills-based experience variables, and explana-
tory variables for hypotheses 1 to 3 (regulatory distance, breadth, and depth),
maintains consistent directional and significance levels of the explanatory
variables. When all the experience types are considered simultaneously in
the full model, the host-country experience and skills-based capabilities expe-
rience variables are not significant in MNCs’ learning curves. This suggests
that repetitive entry with dissimilar intuitional experience negates the bene-
fits of repetition. Model 12 also reveals that positive learning-curve effects
associated with depth are strong enough to negate the negative learning
curve effects of both regulatory distance (b = –.03) and cultural distance
Perkins 169
Log−logistic Regression
.8 .6
Hazard Function
.2 .40
Log−logistic Regression
.8 .6
Hazard Function
.2 .4 0
* Includes all controls and select explanatory variables from model 12.
combined. Among all the models in tables 4a–4c, the two leading explanatory
variables that have the greatest effects on survival are the sign of direction
of regulatory distance, which accelerates failure time by a factor of .44, and
the depth of experience, which decelerates failure time by a factor of .27.
170 Administrative Science Quarterly 59 (2014)
This suggests that the learning gaps created by dissimilar institutional envir-
onments can be overcome to a great extent by recombining related prior
institutional experiences.
Geographic distance could also affect the propensity to invest in certain coun-
tries, which may have an indirect effect on how firms acquire regulatory experi-
ence. To confirm the exclusion restriction requirement (Greene, 2000: 672) of
not being correlated with success/failure, I separately added auction wins and
geographic distance to the probit models in table A13 in the Online Appendix.
Neither was correlated with firm survival.
To address the potential sample selection bias resulting from the heteroge-
neity in firms’ decisions to enter Brazil or not, I used these two valid instru-
ments and the three explanatory variables in the Heckman probit selection
models in table A13 (modified version of Heckman, 1979; adequate techniques
for event-history analysis have not yet been developed) (Boehmke, Morey, and
Shannon, 2006) to test whether the error terms of the two equations were cor-
related (r). To make the coefficients comparable, I replicated model 12 specifi-
cations, fitting a maximum likelihood probit model on the 779 observations
used in the survival analysis, and found similar results to those in model 12. I
constructed three selection models to address all possible unobserved hetero-
geneity concerns. Ideally, selection should include the entire population of mul-
tinational telecom firms that ever won a license that could have entered Brazil
during this period. The first selection model, global entry, examined this entire
population of firms. A more precise approach is to test for selection biases
among the firms that expressed an interest in entering Brazil. This set of firms
was examined in the next model, Brazil entry. The last model used a combined
approach to address both selection and endogeneity identification problems
together. I used a double-selection model (Amemiya, 1985) in which the first
selection equation is conditioned on having regulatory experience or not (0/1
dummy splitting the explanatory variable on the mean), and the second selec-
tion model is conditioned on the decision to enter Brazil or not. I included the
inverse Mill’s ratio from the first selection model in the latter to account for the
endogeneity of regulatory distance. Results across all three selection models
revealed there was no significant selection bias among the entering firms. The
coefficient estimates remain consistent for both the global entry and Brazil
entry models, though the significance of the regulatory distance variable has
weakened in the global entry model. Conditioning the results on both sample
selection and endogeneity only strengthened the magnitude of the coefficients.
Mainly, these selection bias results provide internal validity to the explanatory
power of the hypothesized effects.
Robustness Checks
I conducted 18 separate robustness checks in models 13 to 30; results are
available upon request. Models 13 to 15 tested the institutional control vari-
ables with multicollinearity concerns in lieu of the institutional controls used in
models 2 to 12, including language distance (Grimes, 1992), egalitarianism
(Siegel, Licht, and Schwartz, 2011), and legal origin distance (La Porta et al.,
1998).9 I also tested whether the results are robust during presidential and min-
isterial regime changes (Siegel, 2007) in models 16 and 17. The disaggregated
9
Egalitarianism measures were converted into Euclidian distance scores. Data sources included
Schwartz (1994) and table 1 of Siegel, Licht, and Schwartz (2011) for missing country data.
Germany was used as a proxy for Luxembourg.
172 Administrative Science Quarterly 59 (2014)
Hofstede (1980) cultural distance composite measure was included as the four
dimensions of culture—PDI, IDV, MAS, and UAI—separately (Kogut and Singh,
1988; Shenkar, 2001) in models 18 to 22. Robustness checks in models 23 to
27 provided additional firm-level performance measures, including the number
of subscribers, purchase price, ARPU (average revenue per user), and number
of employees (alternative proxy for firm size). A robustness check in model 28
tested the first entry-year observations only. Two final robustness checks in
models 29 and 30 tested alternative measurements of regulatory distance
including Euclidian distance and standardized regulatory scores found in Online
Appendix F. Results revealed that the regulatory distance variable remains con-
sistently significant (p < .001) in all robustness checks except model 28 (first
entry-year observations). The breadth variable was significant across all robust-
ness checks except in model 14 (legal origins) and model 28. Similarly, the
depth variable was significant in all robustness checks except in models 14
(legal origins), 15 (egalitarianism), and 28. Alternative measurements in models
29 and 30 also produced similar coefficient estimates and significance levels.
Model 31, the final robustness check, examined the results including an eco-
nomic distance measure from Online Appendix F. These 18 robustness checks
indicate mostly consistent and repetitive support for the explanatory variables.
Moreover, the results from the robustness check in model 25 are consistent
with the egalitarian distance measure used in Siegel, Licht, and Schwartz
(2011); Hofstede’s (1980) uncertainty avoidance (UAI) index (used by Kogut and
Singh, 1988) was tested in model 21 and was consistently negative but not
significant.
investment are associated with learning discounts when other similarities are
accounted for is another indication of dissimilar experience as a learning hin-
drance. These finding are counter to the theoretical predictions of Johanson
and Vahlne (1977), who also argued that benefits of reduced market uncer-
tainty should also be derived from investments across successive new coun-
tries. These results suggest that adding country-level experiences that neither
have regulatory similarity nor exploit skills-based capabilities of the firm are tax-
ing to subsequent foreign investments. Perhaps this refinement to the existing
theoretical view is needed. Future studies should take into account the learning
discount when knowledge that may not be relevant to the targeted institutional
environment is transferred across countries.
In exploring types of experience, this study provides a framework to identify
the unique and inimitable regulatory experiences that firms gain from idiosyn-
cratic patterns of sequential foreign investments. One future extension of
this research could explore how such knowledge can be leveraged as a
source of competitive advantage. The challenge for organizations is to recog-
nize the similarities that subunit organizations can experience across key
institutional dimensions and that can subsequently be retrieved in related
knowledge situations. The informed investment manager is able to select
investments that appear to be risky to the inexperienced by deploying unique
institutional knowledge based on related country characteristics (Davidson,
1980). These investment benefits may appear intuitive; however, interviews
with 30 telecommunications executives conducted for this study revealed
that many foreign investment managers do not make such underlying con-
nections and are more frequently blindsided by institutional differences.
Perhaps the heterogeneity in organizations’ regulatory experience provides a
plausible explanation for the unresolved variation in organizational learning
rates. The learning-curve differences among industry competitors could
result from the sequence and frequency of types of experience acquired
across learning environments. Organizations with investment patterns that
reduce the knowledge gaps across investments are likely to learn at a faster
rate than firms with fewer synergies.
This study also presents another key distinction in the variation in institu-
tional environments between countries. The six dimensions of the telecom reg-
ulatory framework explored in this study can be generalized to other regulatory
contexts, including energy, mining, and other utilities, pharmaceutical patent
protection and drug content regulations, state and local health care regulations,
and the banking industry. Subsequent studies assessing cross-country compari-
sons should include such regulatory dimensions in the same tradition as more-
established institutional measures such as culture.
Acknowledgments
I especially thank Bernard Yeung, Rachelle Sampson, Juan Alcacer, Goncxalo Pacheco de
Almeida, Jeffrey Robinson, Edward Zajac, William Ocasio, Paul Hirsch, Klaus Weber,
and Ray Reagans for their helpful suggestions. I also thank participants at the Academy
of Management 2005 Annual Meeting, the Strategic Management Society Annual
Conference in 2004, the Academy of International Business Annual Conference in 2004,
and seminar participants at Kellogg School of Management, University of Michigan,
Columbia Business School, Duke University, Ohio State University, University of
Minnesota, University of Maryland, Georgetown University, University of Southern
California, INSEAD, Rutgers Business School, Texas A&M University, University of
Illinois, NYU Stern School of Business Management and Organizations Department
176 Administrative Science Quarterly 59 (2014)
Brown Bag Seminar, NYU Stern Global Business Institute Cross-Disciplinary Strategy
Seminar, and CCC Colloquium for Doctoral Student Research for their valuable feedback
on an earlier version of this paper. I also acknowledge NYU Stern School of Business for
financial support and am grateful to Sergio Lazzarini, of Insper, São Paulo, for providing
the opportunity and support to conduct field research in Brazil. All errors and omissions
remain my responsibility.
REFERENCES
Allison, P.
1999 Multiple Regression: A Primer. Thousand Oaks, CA: Pine Forge Press.
Almeida, P., and B. Kogut
1999 ‘‘Localization of knowledge and the mobility of engineers in regional networks.’’
Management Science, 45: 905–917.
Amburgey, T. L., and A. S. Miner
1992 ‘‘Strategic momentum: The effects of repetitive, positional, and contextual
momentum on merger activity.’’ Strategic Management Journal, 13: 335–348.
Amemiya, T.
1985 Advanced Econometrics. Cambridge, MA: Harvard University Press.
Argote, L.
1999 Organizational Learning: Creating, Retaining and Transferring Knowledge.
Boston: Kluwer Academic.
Argote, L., S. Beckman, and D. Epple
1990 ‘‘The persistence and transfer of learning in industrial settings.’’ Management
Science, 36: 140–154.
Argote, L., and D. Epple
1990 ‘‘Learning curves in manufacturing.’’ Science, 247: 920–924.
Barkema, H., J. Bell, and J. Pennings
1996 ‘‘Foreign entry, cultural barriers and learning.’’ Strategic Management Journal,
17: 151–166.
Beckman, C. M., and P. R. Haunschild
2002 ‘‘Network learning: The effects of partners’ heterogeneity of experience on
corporate acquisitions.’’ Administrative Science Quarterly, 47: 92–124.
Benito, G., and G. Gripsrud
1992 ‘‘The expansion of foreign direct investment: Discrete rational location choice or
a cultural learning process?’’ Journal of International Business Studies, 23: 461–476.
Berry, H.
2006 ‘‘Shareholder valuation on foreign investment and expansion.’’ Strategic
Management Journal, 27: 1123–1140.
Berry, H., and M. Sakakibara
2008 ‘‘Resource accumulation and overseas expansion by Japanese multinationals.’’
Journal of Economic Behavior and Organization, 65: 277–302.
Boehmke, F. J., D. S. Morey, and M. Shannon
2006 ‘‘Selection bias and continuous-time duration models: Consequences and a
proposed solution.’’ American Journal of Political Science, 50: 192–207.
Brüderl, J., and R. Schüssler
1990 ‘‘Organizational mortality: The liabilities of newness and adolescence.’’ Adminis-
trative Science Quarterly, 35: 530–547.
Carroll, G. R., and Y. Huo
1986 ‘‘Organizational task and institutional environments in ecological perspective:
Findings from the local newspaper industry.’’ American Journal of Sociology, 91:
838–873.
Perkins 177
Caves, R. E.
1996 Multinational Enterprise and Economic Analysis, 2d ed. New York: Cambridge
University Press.
Chang, S.
1995 ‘‘Internationalization expansion strategy of Japanese firms: Capability building
through sequential entry.’’ Academy of Management Journal, 38: 383–407.
Chang, S.
2003 ‘‘Ownership structure, expropriation and performance of group affiliated compa-
nies in Korea.’’ Academy of Management Journal, 46: 238–253.
Cyert, R. M., and J. G. March
1963 A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall.
Darr, E. D., L. Argote, and D. Epple
1995 ‘‘The acquisition, transfer, and depreciation of knowledge in service organiza-
tions: Productivity in franchises.’’ Management Science, 41: 1750–1762.
Davidson, W.
1980 ‘‘The location of foreign direct investment activity: Country characteristics and
experience effects.’’ Journal of International Business Studies, 11: 9–22.
Delios, A., and P. Beamish
2001 ‘‘Survival and profitability: The role of experience and intangible assets in
foreign performance.’’ Academy of Management Journal, 44: 1028–1038.
Delios, A., and W. Henisz
2000 ‘‘Japanese firm investment strategies in emerging economies.’’ Academy of
Management Journal, 43: 305–323.
Delios, A., and W. Henisz
2003 ‘‘Political hazards, experience and sequential entry strategies: The international
expansion of Japanese firms 1980–1998.’’ Strategic Management Journal, 24:
1153–1164.
Demsetz, H.
1968 ‘‘Why regulate utilities?’’ Journal of Law and Economics, 11: 55–65.
DiMaggio, P. J., and W. W. Powell
1983 ‘‘The iron cage revisited: Institutional isomorphism and collective rationality in
organizational fields.’’ American Sociological Review, 48: 147–160.
Dobbin, A., B. Simmons, and G. Garrett
2007 ‘‘The global diffusion of public policies: Social construction, coercion, competi-
tion, or learning?’’ Annual Review of Sociology, 33: 449–472.
Dunning, J.
1980 ‘‘Toward an eclectic theory of international production: Some empirical tests.’’
Journal of International Business Studies, 11: 9–31.
Dutton, J., A. Thomas, and J. Butler
1984 ‘‘The history of progress functions as a managerial technology.’’ Business
History Review, 58: 204–233.
Epple, D., L. Argote, and R. Devadas
1991 ‘‘Organizational learning curves: A method for investigating intra-plant transfer
of knowledge acquired through learning by doing.’’ Organization Science, 2: 58–70.
Epple, D., L. Argote, and K. Murphy
1996 ‘‘An empirical investigation of the microstructure of knowledge acquisition and
transfer through learning by doing.’’ Operations Research, 44: 77–86.
Erramilli, K.
1991 ‘‘The experience factor in foreign market entry behavior of service firms.’’ Jour-
nal of International Business Studies, 22: 479–501.
Evans, P.
1979 Dependent Development: The Alliance of Multinational, State, and Local Capital
in Brazil. Princeton, NJ: Princeton University Press.
178 Administrative Science Quarterly 59 (2014)
Author’s Biography
Susan E. Perkins is an assistant professor in the Management and Organizations
Department and the International Business and Markets Program at the Kellogg School
of Management, Northwestern University, 2001 Sheridan Rd., Evanston, IL 60208
(e-mail: s-perkins@kellogg.northwestern.edu). Her research examines how institutional
variations between nations potentially create investment risk factors for multinational
corporations. Her research focuses on the international business implications of industry
regulation, corporate governance and ownership structure, experiential learning, and
firm-level non-market strategy. She holds an M.B.A., Master’s of Philosophy, and Ph.D.
in international business strategy from the Stern School of Business, New York
University.