A note on bilateral trade agreements in the presence of

irreversible investment and deferred negotiations

Matthew Cole
University of Oregon

M. Ryan Haley Aaron Lowen

University of Wisconsin - Oshkosh Grand Valley State University

Abstract
A common result in the trade literature is that a small country will realize gains from a
bilateral free trade agreement with a large country. McLaren (1997) casts aspersions on this
traditional belief by demonstrating that irreversible investment in the small country, with the
possibility of re-negotiation by the large country, can actually make the small country prefer
autarky to free trade. In this note, we identify a middle ground where the small country can
realize above-autarky utility by only partially specializing (relative to the free-trade level of
specialization) in export production; this improvement occurs even in the presence of
irreversible investment and deferred negotiations.

We are grateful to Ray Riezman and Ron Davies for helpful comments and useful suggestions.
Citation: Cole, Matthew, M. Ryan Haley, and Aaron Lowen, (2008) "A note on bilateral trade agreements in the presence of
irreversible investment and deferred negotiations." Economics Bulletin, Vol. 6, No. 34 pp. 1-10
Submitted: June 25, 2008. Accepted: August 26, 2008.
URL: [Link]
 1. INTRODUCTION

Small countries are often encouraged to embrace trade, especially free trade. Advocates of this
recommendation point to the potential gains from trade (relative to autarky) and the difficulties a
small country may encounter if it attempts protectionist policies.1 By this reasoning, a small
country that does not adhere to free trade is not acting optimally, or so it would appear.2
McLaren (1997) questioned this conventional wisdom: What if the small country’s
investment was irreversible and there were deferred negotiations? Generally, when a small
country enters a trade agreement with a large country, the small country will reapportion its
resources to the production of the good to be exported to the large country. McLaren points out,
however, that if this investment is sunk, the small country will lose its ability to effectively re-
negotiate in subsequent trade talks; i.e., it tools-up specifically for the large country’s needs (at
the expense of its own import sector) in exchange for favorable trade conditions, only to have the
trade conditions undone once tooling is complete. The small country’s myopia precipitates a
need for the large country’s market thereafter, and a corresponding loss of bargaining power for
future trade talks – the so-named “negative strategic externality”. If this erosion in bargaining
power is substantial, the large country can extract concessions or side payments from the small
country, possibly making it worse off than autarky; McLaren discusses some historical examples
in detail.3
Looking at trade through this aperture casts doubt on the conventional belief that trade is
generally a “good deal” for the small country. McLaren’s insights revitalize the debate about the
merits of autarky and protectionism from the small country’s perspective. Indeed, his results
make isolationism look attractive for the small country.
To combat the shortcomings faced by the small country, McLaren (pg. 417) suggests that
its government create a subsidy to maintain its autarky resource allocation:

In addition, under either anticipated cooperation or trade war, a subsidy just large
enough to induce the same allocation of Home [small country] resources as would
obtain under autarchy would also be welfare improving, because then ex post
Home utility would be no lower than under autarchy.

The intent of the subsidy is to prevent small-country agents from zealously investing in the up-
and-coming export sector, hence mitigating the negative strategic externality.4 Even a small
subsidy, McLaren argues, would improve the small country’s ex post utility.5

1
Even the protection of infant industries has been called into question; see, for example, Baldwin (1969).
2
Throughout this note, and in the spirit of McLaren (1997), we use the term “free trade” to mean trade without
tariffs; other modes of protectionism (e.g., subsidies or quotas) are permissible within this free-trade paradigm.
3
While this note works directly from McLaren (1997), other researchers have explored the implications of sunk
investment in the presence of trade of agreements; see, for example, Park (2000) or Bond and Park (2002). Both of
these papers deal with a more dynamic setting with possibilities of punishment. Park (2000) also has temporary
irreversibility whereas McLaren (1997) has permanent irreversibility. In a related study, Chisik and Davies (2004)
consider the impact of irreversible FDI.
4
At first glance it seems that individually-rational, small-country agents should recognize that aggressive
(irreversible) investment in the export sector will undermine their future bargaining power as a nation. However,
McLaren (1997 pg. 404) contends ``... in the problem of national bargaining... decentralized investment decisions
destroy the ``individual rationality’’ constraint...’’

1
 While the myopic free-trade outcome of complete specialization is clearly non-optimal
for the small country facing sunk investment and deferred negotiations, we show that McLaren’s
subsidy recommendation does not maximize the small country’s utility in all circumstances.
Describing an alternative, which prescribes partial specialization and therefore a more modest
level of small-country protectionism, is the purpose of this note.
Specifically, we present a situation where it is optimal for the small country to protect its
import sector and specialize in the export good, though not to the degree implied by free trade.
Maintaining diversified production along with a measured amount of protectionism allows the
small country to enjoy above-autarky utility levels. We demonstrate our approach in a simple
two-period model using two cases, the leading case involving Cobb-Douglas utility and the other
involving CES utility.

2. THE MODEL

Following McLaren (1997), we use a simple two-period Ricardian model wherein two countries,
one small and the other large (demarcated by “*” hereafter), are poised to trade. We assume
production is linear with labor as the only factor, and that each worker supplies one unit of
labor.6 We use L and L* to denote the labor supplies in the respective countries, where L* > L > 0.

  1 units of good X or 1 unit of Y; one unit of labor in the large country can either produce 1
There are two goods, X and Y. A unit of labor in the small country can produce either

unit of X* or Y*; i.e.,

 YL X *  Y *  L* .
X

All consumers are assumed to have identical Cobb-Douglas utility functions:

U (C X , CY )  C X CY .
We assume that the large country may choose to impose a per-unit tariff, which we
denote as τ* .7 We also assume that the small country has various non-tariff protectionist policies
(e.g., quotas or subsidies) at its disposal, though we do not explicitly incorporate them in the
model.8 This assumption does not impact our intended contribution because we are interested
instead in a more general question: Should the small country erect trade barriers or not? We are
deliberately agnostic about the form of protectionism the small country may choose.

5
Here, “ex post” refers to the time after the trade agreement has been re-negotiated.
6
While a more general production technology could be used, our intended contribution builds directly on
McLaren’s formulation. We conjecture that our results would qualitatively hold in the presence of more
sophisticated production technologies provided there is a clear comparative advantage to be exploited in a bilateral
trade environment. For example, a technology (e.g., in a Heckscher-Ohlin framework) exhibiting a bowed-out PPF
would likely reduce the severity of the deferred negotiations (in the case of expected cooperation) because the small
country will not fully specialize. However, once production levels are chosen and fixed, the small country
essentially becomes an endowment economy, as is the case in our model. So, unless the curvature is such that they
will choose autarky in period one, there is likely to be a middle-ground (i.e., a difference between autarky and trade
levels of specialization) where our result can arise.
7
Since the autarky price of X* is 1, an ad valorem tariff produces the same result.
8
The small county has the option to charge a tariff, but its optimal tariff in first-round negotiations is zero.

2
 Our focus is on the case where the small country is very small relative to the large
country, which corresponds with L/L* ≈ 0. An implication of this assumption is that the small
country has no impact on the large country’s prices, resulting in fixed prices for the large
country. Letting Y* be the numeraire good induces the following prices:

PX  P PX*  1
 PX*  *  1  *
PY  1    1 PY*  1
The Walrasian demand functions are as follows:

I I *
CX  C X* 
(   )(1  * ) (   )
I I *
CY  CY* 
(   ) (   )

I  (1  * ) X  Y I *  X *  Y *  * M * .
where

The model unfolds over two periods. Two activities occur in period 1: a) the free trade
agreement is signed and b) labor is irreversibly invested (as in McLaren’s paper) in the two
sectors of the small country. Three activities occur in period 2: a) production output is realized,
b) the large country re-negotiates and starts charging its optimal (revenue-maximizing) tariff, and
c) trading commences.
The nature of the trading will either be cooperative or non-cooperative (i.e., trade war). If
the small country anticipates cooperation it will fully specialize in the production of the good
slated for export to the large country – here good X, as that is where its comparative advantage
lies. This ultimately allows the large country to extract all the utility from the small country
when re-negotiation occurs; i.e., the small country will have irreversibly tooled-up for the export
good, only to have the large country take advantage of it thereafter. If the small country
anticipates a trade war, the rational expectations equilibrium occurs wherein the small country is
indifferent between autarky and trade.
Neither of these outcomes gives the small country an incentive to enter the trade
agreement in period 1: with anticipated cooperation, the small country ends up worse off than
autarky and with an anticipated trade war, the small country would at best be indifferent between
autarky and trade.9 However, if we allow for government intervention, the small country can in
fact realize gains from trade, provided the specialization decision is made carefully.
To see how the small country’s government can beneficially intervene, first let ψ
represent changes in production in the small country, interpreted as units of labor moved from
sector Y to sector X; i.e., ψ gauges the small country’s intended degree of specialization. The
value ψ is chosen in period 1 and determines the labor allocation that is then fixed, because of
the irreversibility assumption, for period 2.

9
The outcome of a trade war depends on elasticities and country size, where bigger is better; see Kennan and
Riezman (1988) or Syropoulos (2002). See Mayer (1981) for a discussion of bargaining power differentials between
countries of asymmetric size.

3
 For each   0, 
L
 there is a particular endowment of X and Walrasian demand for X:

L
X    
 

C X  * ;   
I
(   )(1  * )

 L  L
where

I  (1  * )    .
     

 
E X * ;   X     C X  * ;   ,
Using the small country’s export function

the large country’s optimal tariff obtains by maximizing revenue:

* E X  * ;   ,

max
*

which implies that the large country’s optimal tariff is10

  L   (    ) 
  1 
  L   (   )
*
.

Note that this is increasing in ψ, the degree to which the small country specializes.11
With the large country’s optimal tariff in hand, the government of the small country can
maximize utility by choosing its degree of specialization. A wise small-country government will
seek to limit the degree of specialization (i.e., keep it less than the free-trade level), which in turn
lowers the large country’s optimal tariff. The net result is a “best” amount of specialization for
the small country, which maps to above-autarky levels of utility for the small country – a more
encouraging outcome for the small country than McLaren foresees.
This result is interesting because it falls somewhere between the conclusions of
traditional trade theory and McLaren’s critique. Under the latter, the small country would be, at
best, indifferent between trade and autarky; here, there is a clear reason (i.e., higher utility) to
engage in trade even if investment is irreversible and trade terms are likely to be re-negotiated
(by the large country) to the detriment of the small country. We summarize this finding in the
following proposition.

10
We can now verify McLaren's result that the small country may prefer autarky to a free trade agreement (and

  L /(   ), the numerator equals zero and the denominator is strictly positive. Therefore, the large country’s
deferred negotiations) with the large country. If the small country is myopic and fully specializes, then

optimal tariff equals one, which results in the small country’s price (P) equaling zero, and consequently its income

This follows from    L /(   ).

equals zero as well.
11

4
PROPOSITION. The small country will experience an increase in utility following an increase in
production of its export good relative to autarky production levels, even when the large country
charges its optimal tariff.

 
PROOF. Consider the maximization problem faced by the small country:

max U C X , CY , ψ; τ ,
*

or, more specifically,

     
  (1   * ) L    L    
     
  (1   * ) L    L    
 

   
max     

 (   )(1   )   (   ) 
   
*

  L   (   )
where

  1 
  L   (   )
*
.

Substituting for τ , differentiating, setting ψ = 0, and assuming that γ + β = 1 yields

*

U (•)        2      2    
 

2      
 0.

฀

Because the first partial is positive at ψ = 0, the utility of the small country will increase with an
increase in production of its export good.12 The point where ψ = 0 is of particular interest
because it coincides with the pre-trade allocation of labor – the circumstance in which the small
country will be in when considering the period-1 trade agreement. As we will demonstrate below
for specific parameter values, the partial derivative above decreases but remains positive as ψ
increases, ultimately reaches zero at the point ψ Umax (consider the slope of the utility function
depicted in figure 1), and is negative for values larger than ψ Umax . The value ψ Umax is the point at
which the small country’s utility is maximized, and it occurs, interestingly, somewhere between
the autarky and rational expectation levels of specialization (indicated by ξ).

3. NUMERICAL EXAMPLES

We have demonstrated that the small country can gain (relative to autarky) by specializing
strictly less than the rational expectations equilibrium would indicate; to see this clearly we
provide two numerical illustrations. In figure 1, we depict the outcome when the parameter set
{γ, β, α} equals {0.5, 0.5, 2.5}. As described above, there exists a range of positive values for ψ

12
The constant-returns-to-scale assumption is not crucial; we impose it here to tidy the presentation.

5
where the small country gains from trade and partial specialization. Movement along the
horizontal axis indicates the small country shifts labor from import sector Y to export sector X.
At each ψ there is a new export function and consequently a new optimal tariff. The point of
specialization where with-trade utility equals autarky utility is labeled ξ, and is the unique
rational expectations equilibrium. If the small country specializes anywhere from zero to this
point, it will be at least as well off as in autarky; ψ Umax represents the level of specialization that
maximizes utility.
If we hold the utility parameters (γ and β) constant and increase α, we can see how the
landscape changes when the small country is more efficient at producing good X. As expected,
the autarky utility level increases, as do the gains from trade with autarky levels of production.
The optimal level of specialization ψ Umax increases as well. In table I, we summarize three
additional representative parameterizations (we experimented with others as well) for the Cobb-
Douglas model; for purposes of comparison, values of specialization ( ψ ) have been normalized
to be a percent of total labor allocated to autarky production of the imported good. In all
instances, partially specialized production offers the small country higher utility than the autarky
resource allocation. Thus, a small country facing trade negotiations involving irreversible
investment and deferred negotiations (as posed by McLaren, 1997) may prefer a smaller subsidy
(or equivalent) than McLaren recommends.
Table II contains numerical examples for the more general CES utility function, showing
the same qualitative results. For these and other values (not reported here) we find, as in the
Cobb-Douglas case, that the small country facing trade negotiations involving irreversible
investment and deferred negotiations may prefer a smaller subsidy (or equivalent) than McLaren
recommends.

4. CONCLUSIONS

It is commonly believed that small countries greatly benefit from free trade. McLaren (1997)
argued to the contrary by demonstrating that, in the presence of irreversible investment and
deferred negotiations, a small country may actually have very little to gain from trading and that
it should consider using protectionist policies. McLaren (1997) recommended a subsidy that
would keep the small country’s resource allocation fixed at autarky levels, thus avoiding the
deleterious effects of the negative strategic externality.
In this note we demonstrate that targeting the autarky resource allocation may, in fact, not
be the small country’s utility-maximizing course of action in the presence of irreversible
investment and deferred negotiations. We find that the small country’s best strategy is to
partially specialize in the export good, and that the small country should accomplish this using
some form of protectionism, though we are deliberately agnostic about which instrument should
be used. The implications of this note should be of interest to trade-policy makers in small and
large countries alike.
Our contribution can also be understood in a somewhat broader context. For example,
adding more countries to the model has some interesting implications, which we outline below in
three cases.

Case 1: Suppose one large country signs a bilateral free trade agreement with
multiple, identical small countries. By the definition of a small country, each
country cannot affect the world relative price. In this vein, one could think of each

6
 country as decentralized firms and both the anticipated cooperation and rational
expectations results hold. If we allow for a benevolent government (as in our
model), it does not appear obvious that the “optimal” level of specialization for
each country would be any different than presented in our model, assuming there
is no collusion on the part of the small countries. If we allow the small countries
to form a trade union, such as a customs union, then the optimal strategy of the
customs union will depend on the collective size of the union; see, for example,
Abrego, Riezman, and Whalley (2005).

Case 2: Suppose a bilateral free trade agreement is signed between multiple large
countries and one or more small countries. Since the optimal tariff of a large
country is the revenue-maximizing tariff, and because there is now competition
for revenue, each large country would lower its tariff to attract the small country’s
exports; this would continue until there was free trade. This scenario assumes zero
(or identical) transaction costs between a small country and all large countries. If
we allow for differing transportation costs, the ability of a large country to re-
negotiate with the small country in the second period would be bounded by these
differing costs.

Case 3: Suppose a bilateral free trade agreement is signed between two WTO
members. Here, the principle of non-discrimination, as embodied in the most-
favored-nation (MFN) clause, plays a key role (see Bagwell and Staiger, 1999). In
this case, the ability of a large country to re-negotiate is bounded by the MFN rate
(assuming there is another member country also exporting the same good).

In the end, however, the synopsis is this: McLaren recommends autarky levels of production
when the potential repercussions from irreversible investment are virtually unbounded – a worst-
case scenario. We show that, even under this worst-case scenario, some specialization can be
optimal for the small country. When the repercussions are bounded by real-world trade features,
like those described in cases 2 and 3, McLaren’s recommendation of autarky levels of production
is more questionable because the damage the large country can do after re-negotiation is now
limited; i.e., the binding of the large country's second-period tariff increases the small country's
optimal level of specialization. Because our argument for some specialization obtains in the
circumstance that is most hostile towards doing so, adding institutional realities (e.g., cases 2 and
3) strengthens our contention that McLaren's prescription of autarky production is too strong.

7
 Figure 1
Small Country’s Utility Given Level of Specialization

 = 2.5;  = 0.5;  = 0.5

Utility

Autarky
Utility

Utility w/
Specialization

Umax  
Full
Specialization

Table I: Small Country Cobb-Douglas Parameterizations

Autarky Gains from trade under...
Autarky Specialized
α γ β Utility ψ Umax ξ Production Production
5 0.5 0.5 55.902 24.08% 66.67% 8.20% 10.75%
2.5 0.5 0.5 39.526 14.71% 42.86% 2.64% 3.49%
2.5 0.25 0.75 35.829 8.88% 27.27% 2.13% 8.88%
2.5 0.75 0.25 56.651 18.90% 52.94% 1.83% 2.34%

Table II: Small Country CES Parameterizations

Autarky Gains from trade under
α σ Utility ψ Umax ξ Autarky Specialization
5 0.5 23.8729 8.39% 27.64% 3.94% 5.24%
2.5 0.25 29.0426 2.71% 8.27% 0.64% 0.85%
2.5 0.5 18.7624 5.88% 18.45% 1.30% 1.74%
2.5 0.75 4.8134 9.34% 29.98% 1.98% 2.62%

8
 5. REFERENCES

Abrego, L., Riezman, R. and J. Whalley. (2005) “Computation and the Theory of Customs
Unions” CESifo Economic Studies, 51,117-132.

Bagwell, K. and R. Staiger. (1999) “An Economic Theory of GATT” American Economic
Review 89, 215-248.

Baldwin, R. (1969) “The Case against Infant-Industry Tariff Protection” Journal of Political
Economy 77, 295-305.

Bond, E. and J. Park. (2002) “Gradualism in Trade Agreements with Asymmetric Countries”
Review of Economic Studies 69, 379-406.

Chisik R, and R. Davies. (2004) “Gradualism in Tax Treaties with Irreversible Foreign Direct
Investment” International Economic Review 45, 113-139.

Kennan, J. and R. Riezman. (1988) “Do Big Countries Win Tariff Wars?” International
Economic Review 29, 81-85.

Mayer, W. (1981) “Theoretical Considerations on Negotiated Tariff Agreements” Oxford

Economic Papers 29, 135-153.

McLaren, J. (1997) “Size, Sunk Costs, and Judge Bowker's Objection to Free Trade” American
Economic Review 87, 400-420.

Park, J.H. (2000) “International Trade Agreements Between Countries of Asymmetric Size”
Journal of International Economics 50, 473-495.

Syropoulos, C. (2002) “Optimum Tariffs and Retaliation Revisited: How Country Size Matters”
Review of Economic Studies 69, 707-727.