Intersectoral conflict and delays in macroeconomic stabilization
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Razmi, Arslan Working Paper Intersectoral conflict and delays in macroeconomic stabilization Working Paper, No. 2024-4 Provided in Cooperation with: Department of Economics, University of Massachusetts Suggested Citation: Razmi, Arslan (2024) : Intersectoral conflict and delays in macroeconomic stabilization, Working Paper, No. 2024-4, University of Massachusetts, Department of Economics, Amherst, MA, https://hdl.handle.net/20.500.14394/22344 This Version is available at: https://hdl.handle.net/10419/300486 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/
University of Massachusetts Amherst University of Massachusetts Amherst ScholarWorks@UMass Amherst ScholarWorks@UMass Amherst Economics Department Working Paper Series Economics 2024 Intersectoral Conflict and Delays in Macroeconomic Stabilization Intersectoral Conflict and Delays in Macroeconomic Stabilization Arslan Razmi Economics Department, University of Massachusetts Amherst Follow this and additional works at: https://scholarworks.umass.edu/econ_workingpaper Part of the Macroeconomics Commons Recommended Citation Recommended Citation Razmi, Arslan, "Intersectoral Conflict and Delays in Macroeconomic Stabilization" (2024). Economics Department Working Paper Series . 356. Retrieved from https://scholarworks.umass.edu/econ_workingpaper/356 This Article is brought to you for free and open access by the Economics at ScholarWorks@UMass Amherst. It has been accepted for inclusion in Economics Department Working Paper Series by an authorized administrator of ScholarWorks@UMass Amherst. For more information, please contact [email protected].
Intersectoral Con‡ict and Delays in Macroeconomic Stabilization Arslan Razmi March 22, 2024 Abstract An important body of literature explores the political economy reasons underlying delays in macroeconomic stabilization. This paper develops a framework to analyze con‡ict between two groups of economic actors, one that has an endowment of internationally tradable goods and another that is endowed with non-tradable goods. The focus is on the exchange rate policy in a developing country set-up where the government employs seigniorage revenue to …nance spending pre-stabilization, and faces …scal and balance of payments problems that necessitate stablization with a step devaluation. The presence of exchange rate and endowment uncertainty, the role of foward-looking expectations, and the possibility of IMF aid in‡uence the likelihood, timing, and terms of a national consensus on stabilization in interesting ways. JEL classi…cation: E31, F34, F41 Keywords: Macroeconomic stabilization, seigniorage, in‡ation, devaluation, capital ‡ight, IMF programs. Department of Economics, University of Massachusetts, Amherst, MA 01003; email: [email protected] 1
When the [villager] was caught stealing onions, the Panchayat [village council] gave him the option of choosing between two punishments: eating a 100 onions or su¤ering a 100 lashes. He chose onions, gave up after eating 10 and asked for lashes; then changed his mind after receiving 10 lashes, and so on. Ultimately, he ended up bearing the agony of both. –Folk story from northern India.1 1 Introduction On December 13, 2023 the Argentine government announced a 50 percent devaluation of the peso against the US dollar (see Figure 1). This was after a long period of relatively gentle but steady depreciation of the peso against major international currencies, high and increasing in‡ation, and large …scal and current account de…cits. A similar if slightly less dramatic pattern played out thousands of miles away in Pakistan as the country negotiated terms for a new IMF stabilization and adjustment program. Many if not most economists would argue that these devaluations (and other accompanying measures) were long overdue. More broadly, macroeconomic stabilization programs in developing countries typically follow sustained periods of expanding …scal and current account de…cits. These de…cits, in turn, are often manifestations of deeper underlying macroeconomic problems such as an inability to collect tax revenues, inappropriate aggregate demand policies that lead to exchange rate overvaluation, and a resort to seigniorage to …nance government spending in the face of declining demand for domestic money. The question, however, is why don’t these stabilizations occur sooner than they typically do, especially once the necessity of stabilization becomes obvious to important segments of the economy? Why do some actors, in other words, delay decisive action only to eventually end up, like the villager in the proverbial story, receiving both onions and lashes (i.e., bearing the costs of delay and …nally agreeing on less favorable terms). This paper contributes to the literature addressing this question by analyzing the political economy of delays in stabilization from an under-explored angle, i.e., con‡ict over the desirable post-stabilization level of the exchange rate and the expectations and uncertainty involved therein. This paper is motivated by several observed empirical regularities: The abandonment of a …xed exchange rate regime or crawling peg often accompanies macroeconomic stabilization in developing countries (see, for example, Cornia (2020) for a detailed discussion). Further, this abandonment often involves devaluing the currency to bring its value closer to the open/parallel market rate. As noted above, this meant a 50 percent devaluation in the case of Argentina where, according to reports, the 1This version is cited for its brevity from the October 7, 2013 edition of India Today (https://www.indiatoday.in/magazine/cover-story/story/19900131-militant-movement- holds-kashmir-in-a-state-of-violent-siege-separatism-gets-new-legitimacy-812287-1990-01-30).
0 200 400 600 800 Jan-20 Jan-21 Jan-22 Jan-23 Jan-24 Month-year Figure 1: Argentinian peso per US dollar (monthly rate). Source: The Central Bank of the Argentine Republic ((https://www.bcra.gob.ar/PublicacionesEstadisticas/Principales_variables_datos_i.asp)). gap between the two rates was as high as 150 percent at the time of the devaluation. IMF-supported programs have typically involved a signi…cant nominal depreciation against the US dollar. Often this is motivated by IMF conditionalities that require uni…cation of the o¢ cial and uno¢ cial exchange rate. Gündüz and Darius (2021) report, for example, that the bilateral exchange rate vis-a-vis the U.S. dollar depreciated, on average, by about 13 percent over the period between 6 months prior to and 36 months after IMF program approval.2This issue was also salient during recent IMF negotiations with countries such as Argentina and Pakistan. Devaluations are unpopular among important segments of society since these result in income redistribution. More than half a century ago, Cooper (1971) termed devaluations “one of the most dramatic – even traumatic –measures of economic policy that a government may undertake.” A vivid illustration of this trauma is the calculation by Steinberg and Malhotra (2014) that military dictators lost power during 17 percent of their 48 devaluation episodes and democratic leaders in 38 percent of their devaluations between 1973 and 2006. In particular, producers in the non-tradable goods producing sectors oppose devaluations. Broz et al. (2008) use …rm level data from the World Bank’s World Business Environment Survey and …nd that owners and managers of …rms producing tradable (manufacturing) goods are more 2See also Krueger et al. (2003) and Agenor (2008, chap. 11). 3
likely to report an exchange rate appreciation as a business concern compared to producers of non-tradables (construction and services). Based on OLS and GMM estimates of a baseline panel of 68 developing and 39 developed countries between 1989-2013, Ugurlu and Razmi (2023) …nd that a higher share of non-tradable sector output is correlated with less undervalued RERs. This …nding is consistent with the view that undervalued RER policies are unpopular among non-tradable industries There is some evidence that this dislike for devaluations increases with the share of imported intermediate costs in production. Again, (Ugurlu and Razmi (2023)) provide some econometric evidence in this regard. The …rm-level empirical evidence presented by Egan (2017) supports the view that …rms with a high dependence on imported inputs are unhappy with real depreciation. Interestingly, as noted by Steinberg (2015), even the producers of highly tradable goods sometimes oppose devaluations (or, the case of ‡exible exchange rates, undervaluation), if their imported intermediate content is high. Consistent with these …ndings, Weldzius (2021) argues that an increase in tradable inputs stemming from the globalization of production networks has lowered support for undervaluation even in countries that have intervened in foreign exchange markets to undervalue their currencies. There is signi…cant uncertainty about not just the magnitude of e¤ects that a given degree of exchange rate change will induce but also the degree to which an o¢ cial announcement of change actually pins down the market exchange rate. This uncertainty, in turn, tends to in‡uence how fervent proponents and opponents of stabilization are in their attitudes. See McNamara (2001) for a more detailed discussion. Finally, theory and evidence suggest that expectations of devaluation that result from high in‡ation increase trade misinvoicing and other forms of evasion of taxes on revenues from tradable production (Patnaik et al. (2012)). Trade misinvoicing, in other words, is endogenous to expected exchange rate policy. This paper analyzes the political economy of stabilization with the level of the exchange rate as the central focus. This is important since, apart from being one of the most important macroeconomic prices in the economy, the exchange rate is also di¤erent as a variable from tax rates or government spending since it is directly a¤ected ahead of the time of actual policy change by expectational changes. This focus on the exchange rate and its interplay with the extent of dependence on imported inputs helps shed interesting new light on the possible reasons often underlying delays. The analysis also demonstrates the ambiguous role of foreign aid and aid-related conditionalities –in particular those a¤ecting future exchange rate levels –in facilitating or inhibiting delay. The work here is related to several strands of literature including those analyzing: (1) the role of expectations and instrument uncertainty in in‡uencing 4
economic agents with diverse preferences over a given choice set, (2) the political economy of delayed stabilizations, (3) the preference among large sectors of population for overvaluation, (4) the preference for undervaluation among tradable sector agents, (5) the opposition to undervaluation sometimes even among manufacturers, (6) the political economy of currency crises and IMF stabilization programs. The next section provides some more background. Section 3 develops the formal analytical framework starting with the baseline case where no foreign/IMF aid is involved, then incorporates the possibility of IMF aid in “bad” times, and …nally analyzes the case where pre-announced IMF recommendations take away uncertainty about the future, post-stabilization exchange rate. Section 4 concludes. 2 Background and Literature One could argue, with a lot of justi…cation, that the delay in macroeconomic stabilization commonly experienced in countries on the verge of …scal and balance of payments crises leaves almost everyone worse o¤. This poses interesting questions to ponder for social scientists. Not surprisingly, then, interest in the political economy of macroeconomic stabilization surged during the 1990s following the limited success of numerous programs across the developing world. Drazen (2000) provides a comprehensive, if dated, survey of the initial literature. Among models that explicitly incorporate temporal dynamics to address this question, Alesina and Drazen (1991) provide an early treatment. In their model, rising …scal de…cits require a change in the policy regime which is delayed by an ongoing “war of attrition”between two sides. Each side prefers to shift the bulk of the adjustment costs to the other side but has imperfect information about the latter’s cost structure. Delay results as each side hopes to wait it out. Casella and Eichengreen (1996) extend the Alesina and Drazen model to incorporate expectations of foreign aid, and show that an announcement of aid that lowers the expected burden of adjustment for private groups has ambiguous e¤ects on the likelihood of delay. The lower cost of adjustment facilitates expedition. On the other hand, if there is a lag between the announcement of the aid and its disbursement, this will tend to delay stabilization. With an ongoing war of attrition, an early announcement serves the purpose of accelerating information dispersal, encouraging the loser to concede sooner. Lags between announcements and dispersals, by contrast, present incentives to delay admission of defeat until the arrival of aid. Thus the role of aid depends on the timing of the transfer. The previous two papers assumed an exogenous …scal de…cit. Velasco (1999) changes this in a dynamic context. The bene…ts of the …scal spending are groupspeci…c while the costs are jointly borne. This distorts incentives, leading to overspending and debt accumulation. 5
Drazen (1996) provides a useful big picture taxonomy by pointing out that there are two broad approaches to analyzing delays in macroeconomic policy change and stabilization, each based in a di¤erent manner on heterogeneity of interests. One approach focuses on imperfect information about the net bene…ts that the other group will receive post-stabilization, although the net bene…ts of reform they themselves would receive are known to each group beforehand. This lack of information creates uncertainty about the willingness of the former to pay for reforms which, in turn, means that reforms may be shunned even if they would bene…t a majority if undertaken. The other approach centers around the status quo bias created by uncertainty about the bene…ts of reforms that individual groups will themselves receive. The present contribution falls in the second category, although I focus on the con‡ict caused by the heterogenous e¤ects of exchange rate changes for di¤erent groups. In addition to its practical relevance as a macroeconomic variable – exchange rate and balance of payments issues have been at the core of many a round of IMF negotiations and macroeconomic stabilization programs –focusing on it also enables me to incorporate the role of expectations in a qualitatively di¤erent manner. This comes out starkly in section 3.3 where knowledge about the post-stabilization level of the exchange rate in‡uences the value of the open market exchange rate today. The two papers closest to my framework are Alfaro (2002) and Laban and Sturzenegger (1994). Laban and Sturzenegger (1994) model delay as the rational outcome of distributional con‡ict between two risk-averse groups, the “rich”and the “poor”in a two-period framework. Stabilization involves transfer of resources from the former to the government for onward transfer to the latter. Also, the former have access to technology that allows them to hide income through …nancial adaptation; a skill that develops over time. If poststabilization payo¤ uncertainty is su¢ ciently high, the two groups may end up delaying adjustment even though the relative position of the poor deteriorates over time in the absence of an agreement. This is true even though, unlike in the war of attrition framework, the identity of the side that loses more from undervaluation is known ex-ante to all, and seems to be more in line with the experiences of developing countries such as Argentina and Pakistan discussed brie‡y earlier. I develop a two-period framework to argue, in the spirit of Laban and Sturzenegger (1994), that delays in undertaking much-needed and much-anticipated exchange rate reforms are not best understood as a planner attempting to maximize the welfare of a representative individual. Instead inter-group con‡ict often plays the lead role. The above-mentioned paper directly inspires my model and the logical structure followed by my argument, even though the con‡ict in my case is between two groups that have endowments that di¤er in a di¤erent dimension. One group has an endowment that can be traded freely international markets a while the other has a non-tradable endowment. Moreover, there is no direct tax transfer from one group to the other. The endogenous variable of interest in my case is the nominal exchange rate rather than the tax rate. These changes are motivated by my interest in the scenario where the country faces 6
serious problems on both the …scal and balance of payments fronts, and wishes to reform by shifting to tax collection rather than seigniorage-based …nancing of de…cits. There is a con‡ict of interest between the tradable sector (which experiences a boost in income following a devaluation) and the non-tradable sector (which experiences a loss of income), and this leads to delay in stabilization if the two groups are su¢ ciently risk-averse. The focus on the nominal exchange rate and the balance of payments adds to the existing literature in important ways. First, it is of signi…cant practical importance. The con‡ict of interests between the tradable and nontradable sectors, lies at the heart of Frieden (1991)’s seminal “interest group theory of exchange rate preferences”. Second, focusing on the exchange rate instead of taxes sheds light on additional new aspects of the issue since, unlike taxes: (1) the level of the exchange rate is less determinate; there often exist gaps between the o¢ cial and open/black markets exchange rates – the latter being more widely accessible – even under …xed exchange rate regimes in developing countries facing external account problems, and (2) changing expectations can modify the exchange rate relevant to market participants even before a change in policy is actually implemented. The other paper close to the present one, Alfaro (2002), has a similar set-up to the extent that it analyzes an economy with a tradable sector and a nontradable one. However, the paper signi…cantly diverges in that it employs a cashin-advance set-up and analyzes the political economy of exchange rate based stabilizations. In contrast to the case here, that paper analyzes a temporary stabilization that involves a temporary reduction in the rate of devaluation and a resulting real appreciation to control in‡ation. This policy hurts the tradable goods owners and has ambiguous e¤ects on the non-tradable good owners. In the present paper, a permanent devaluation hurts the relative position of nontradable producers, who may still agree to stabilization under certain conditions, albeit with a delay that further weakens their relative position. Finally, as mentioned earlier, Casella and Eichengreen (1996) analyze the role of the timing of the announcement of foreign aid in in‡uencing the likelihood of delay in stabilization. However, unlike the present contribution, they do so in a continuous time “war of attrition”set-up. A major implication is that aid reduces the burden of adjustment on the group that is the …rst to concede, which expedites concession. In the present set-up, each group has perfect information about the cost structure of the other group so that early announcement does not resolve informational ambiguities and does not induce an earlier concession. Also, their focus is on taxes rather than the exchange rate so that the policy instrument (the tax rate) is completely non-stochastic. 3 Intersectoral Con‡ict: The Model and Analytics As in Laban and Sturzenegger (1994), consider a two-period set-up but with an economy consisting of two types of agents, one called T(for owning a tradable 7
Intuitively, recall that a devaluation reduces N-agent consumption by eN while increasing T-agent consumption by (1 )eT. Thus, the e¤ect of the elimination of in relative pre-stabilization distortionary tax burdens depends on the share of imported inputs. The higher this share, the greater the relative damage to N-agents from a devaluation, and hence the less likely it is that they will agree to stabilize. Corollary 2: Provided that: (a) ("2) + (1 2)g > eN, i.e., the distortionary tax burden pre-stabilization exceeds the post-stabilization tax costs to the N- sector, and (b) S2>1, i.e., a devaluation follows an agreement, a higher share of imported inputs shrinks the possible set of agreements in Period 2 for a given risk premium. Proof: Perhaps the best way to show this is by recalling that policy makers make devaluation a precondition for stabilization in light of the balance of payments situation, and by then considering (12) written in a slightly di¤erent form: R < 1 + ("2) + (1 2)geN eN R+ 1 ("2) + 2g+c(h 2; K)eT (1 )eT (13) Given a step devaluation as a necessary background condition, the term in the square brackets on the right hand side –which is also the right hand side of inequality (11) –must by necessity be greater than one. Since R > 0, this means in turn that ("2) + (1 2)g > eNis a necessary condition for stabilization. Given the satisfaction of this condition (and that S2>1), the second term on the right hand side of (13) implies that a higher share of imported inputs reduces the likelihood of an agreement. N-agents are hurt by a devaluation in proportion to the intermediate share of output. If the share of imported inputs is high enough, the costs of a step devaluation for them outweigh the costs from in‡ation in the absence of stabilization. For later reference, let’s denote the exchange rate agreed on for stabilization in Period 2 by S 2(>1). Thus, for stabilization to occur in Period 2, S2N> S 2>S2T>1. Proposition 3: Given that "2> "1, and 2< 1from Proposition 1, the relative equilibrium position of N-agents in terms of the exchange rate deteriorates over time (i.e., S 2> S 1). Proof: The proof follows from the result that positive …nancial adaptation (greater misinvoicing) in Period 1 results in greater in‡ation in order to …nance the higher tax burden in Period 2. To understand this, consider that N-agents will be willing to stabilize in Period 1 rather than Period 2 if their discounted expected utility from doing so is greater. That is, 14
(1 + )Uf[1 (S1N+R)]eNg U[(1 )eN("1)(1 1)g] +U f[(1 (S 2+R)]eN Let S1Nbe the maximum jump devaluation that N-agents are willing to accept in Period 1 given that there will be stabilization in Period 2. Then the expression above will hold with equality. (1 + )U1 S1N+ReN=U[(1 )eN("1)(1 1)g] +U f[1 (S 2+R)]eNg(14) Inequality (10) tells us that: 1 S2N+ReN= (1 )eN("2)(1 2)g Using this equation and the results from Proposition 1, namely, "2> "1, and 2< 1, allows us to rewrite and simplify expression (14): (1 + )U1 S1N+ReN>(1 + )U[1 (S 2+R)]eN or, S1N< S 2 and, since S1Nis the maximum devaluation acceptable to N-agents, S 1< S1N, so that S 1< S 2(15) With time, the T-agents are able to conceal a greater proportion of their income abroad so that, barring stabilization, their share of the distortionary tax burden is lower in Period 2, improving their bargaining position. Proposition 4: If 2R > (1 )+("1)+(11)g eN+("1)+1g+c(h 1; H1) (1 )eT,it is an equilibrium strategy to delay agreement even as the N-agents lose bargaining power over time and there is a non-empty set of possible agreements in Period 2. Proof: Again, let S1Nbe the implied devaluation that leaves N-agents indi¤erent between agreeing or not to stabilization in Period 1, given that stabilization is desirable in Period 2. Then S1Nis the solution to: (1 + )U1 S1N+ReN= Uf(1 )eN("1)(1 1)gg+U f[1 (S 2+R)]eNg(16) 15
Similarly, the implied devaluation that leaves T-agents indi¤erent, S1Tis give by: (1 + )Uf[(1 ) (S1TR)]eTg= U(1 )eT("1)1gc(h1; H1)+U f[(1 ) (S 2R)]eTg (17) Linearizing after assuming that, for each group, the di¤erence between net endowments is small under the di¤erent tax regimes, and then solving for the two exchange rates, yields, S1N1 1 + R+ 1 +("1) + (1 1)g eN +S 2(18) S1T1 1 + R+1+ 1 ("1) + 1g+c(h 1; H1) (1 )eT +S 2(19) and recognizing that a delay in stabilization requires that S1N< S1T, yields the following condition for delay to be an equilibrium strategy: 2R > (1 )+("1) + (1 1)g eN +("1) + 1g+c(h 1; H1) (1 )eT (20) Again, the degree of exchange rate uncertainty plays a key role in delay. In the case of low uncertainty, delay is unlikely. Intuitively, forward-looking N-agents realize that their bargaining position deteriorates over time in the absence of stabilization and, therefore, have an incentive to reach an agreement. With high uncertainty about the post-stabilization exchange rate, however, the costs of in‡ation are dominated by the potential cost of agreeing to a devaluation which raises imported input costs. How does the share of imported inputs in‡uence the likelihood of delay? Proposition 5: As long as St>1, that is, stabilization involves a devaluation, a higher imported input share increases the likelihood of delay in Period 1 if ("1) + (1 1)g > eN(the costs of in‡ation exceed the post-stabilization tax burden for N-agents). Proof: Expression (20) can be rewritten as: R > 1 + ("1) + (1 1)geN eN +S 2(R+1+eT("2) + 1g+c(h 1; H1) (1 )eT +S 2) (21) From (19), we know that the expression in the curly parentheses must be greater than 1 + as long as there is a devaluation (St>1). A look at the 16
expression in the …rst set of square parentheses on the right hand side reveals that, as long as ("1) + (1 1)g > eN, this term is greater than 1. This, in turn, means that a higher value of renders the satisfaction of (21) more likely. Expressions (18) and (19) tell us that an increase in lowers the level of S at which both parties are indi¤erent if pre-stabilization in‡ation costs exceed their post-stabilization tax burden, and reduce that level in the opposite case. Intuitively, higher import intensity raises the cost of a devaluation for N-agents relative to the distortionary costs of in‡ation, making them more likely to delay agreement. To see the role of uncertainty starkly, consider the case where the burden of post-stabilization taxes on N-agents exactly equals that of the pre-stabilization distortionary costs (i.e., eT=("2) + 1g+c(h 1; H1)). Then, if R= 0, expression (21) tells us that there is no delay in the presence of a devaluation as long as ("1) + (1 1)g > eN. If stabilization involves a devaluation, delay will not happen unless there is exchange rate uncertainty. To summarize the results from this sub-section, in our baseline set-up, a high level of uncertainty will lead to delay in Period 1 and agreement on a level of the exchange rate in Period 2 in the presence of high in‡ation if the uncertainty associated with the level of the exchange rate is high enough to outweigh the costs of in‡ation in Period 1 but not su¢ ciently high to outweigh those costs in Period 2. The ability of the T-agents to adapt to the in‡ationary tax underlies the dynamics of the model. Increased imported input use lowers the ceiling for the exchange rate that is acceptable to the N-agents, making delay more likely at a given level of uncertainty. In the presence of a step devaluation, and relatively high in‡ation costs, a non-zero risk premium is a necessary condition for a delay.8 As mentioned in earlier sections, IMF programs often accompany macroeconomic adjustment in developing countries. Such programs, moreover, typically come with an agreement in favor of a large devaluation to address balance of payment concerns. The next section explores these important aspects. 3.2 Income uncertainty and the possibility of foreign aid The last subsection focused on the role of uncertainty and imported inputs. I now add two new elements: (1) the possibility of foreign (say IMF) aid following a stabilization agreement, and (2) an expected possible recession (decline in endowment) in Period 2. Many countries eventually opt for an IMF adjustment program once they are at the precipice of a major downturn. Suppose that there is a probability of a symmetrical fall in endowments in Period 2. That is, the expected endowments 8Notice that, as in Laban and Sturzenegger (1994), a social planner who weighs the utility of both groups will have no incentive to pursue …nancial adaptation since it simply increases the rate of distortionary in‡ation. 17
in that period are given by: e2i=ei+ (1 )eiAei;i=N; T (22) where < 1and Ai+(1)represents the expected endowment in Period 2. Moreover, suppose that if there is a recession (i.e., things are bad), and if stabilization is agreed upon between the two groups, then IMF aid arrives. Put di¤erently, agents negotiate the exchange rate knowing that IMF aid will arrive following an agreement and conditional on the occurrence of a recession. This aid uniformly reduces the post-stabilization tax burden by a factor 1(where 0< < 1). Thus, if there is no recession, (4) will continue to apply. Otherwise, in Period 2, with IMF aid, gt=(eN+eT) Thus, the expected e¤ective tax rate multiple conditional on agreement being reached in Period 2 is given by: B= [1 (1 )] Note that, 0<B<A<1and keep in mind that 6= 1 iif < 1. Nothing else changes relative to the baseline set-up and the two sets of agents negotiate over the level of the exchange rate following stabilization, and in an environment of uncertainty regarding the post-stabilization value of the exchange rate. In the next sub-section, we will stipulate that the IMF dictates the level of the nominal exchange rate which is pre-announced and known to everyone. It can be shown to still be the case that in‡ation increases while the prestabilization distortionary tax share of the N-agents increases over time; nothing changes here. Moreover, Proposition 3 from the previous subsection continues to hold with the di¤erence that the equilibrium position of N-agents in terms of the exchange rate now deteriorates even more from Period 1 to Period 2 (see the Appendix for the proof). As shown below, for there to be a non-empty set of agreements in Period 2 (see Proposition 2 in the previous subsection), the respective exchange rates and the aggregate condition are now somewhat di¤erent. All of which leads to our next two propositions. Proposition 6: The possibility of foreign aid in Period 2 (i.e. < 1): (i) expands the possible set of values of Swhich allow an agreement in Period 2, and (ii) provides the necessary medium through which expectations of a recession ( i.e., < 1)allow a similar expansion of the possible set of values of S,and (iii) increases the likelihood of an agreement at a given level of R. Proof: To prove that the possibility of foreign aid expands room for agreement in Period 2, consider that the modi…ed conditions for such an agreement corresponding to (10), (11), and (12) are as follows. For N-agents: Uf[1 (S2N+R)AB]eNg> Uf(1 )AeN("2)(1 2)gg 18
)S2N R+ 1 B A+("2) + (1 2)g AeN (23) For T-agents, Uf[(1 )(S2TR)AB]eTg>U(1 )AeTc(h2; H2)("2)2g )S2TR+1+ 1 B A("2) + 2g+c(h 2; K) (1 )AeT (24) and the aggregate condition corresponding to (12) becomes: 2R (1 ) B A+("2) + (1 2)g eN +("2) + 2g+c(h 2; K) (1 )eT (25) The di¤erence from the corresponding expressions (10), (11), and (12) in the previous subsection is the presence of the term B=A (<1) on the right hand sides. The possibility of IMF aid in the case of an agreement raises the maximum level of Sthat N-agents are willing to allow while reducing the minimum level acceptable to T-agents. This expands the set of possible values of the exchange rate over which an agreement can be reached, making stabilization in Period 2 more likely, and proving (i). To establish (ii), notice that, once the possibility of foreign aid is removed, i.e., = 1, then B=A = 1, which yields the same expression as the corresponding expression (12) in the previous subsection where foreign aid was absent. Thus, expectation of a recession does not in itself in‡uence the likelihood of agreement in Period 2 unless < 1 Finally, the presence of the term B=A (<1) in (25), but not in (12), proves (iii). Intuitively, the probability of a recession, does not by itself a¤ect the values of the exchange rate that leave the two sets of agents indi¤erent unless foreign aid is conditional on the occurrence of that recession. This means that the two sides relax their bargaining positions in Period 2 if and only if foreign aid (which lowers their tax burdens) is a possibility. Do the two sides react similarly in Period 1 by relaxing their bargaining positions in response to the possibility of foreign aid conditional on a recession in Period 2? Proposition 7: Given that foreign aid is conditional on a recession, i.e., 6= 1 iif < 1,the possibility of a recession in Period 2 (i.e., < 1) increases the likelihood, at a given level of R,that an agreement to stabilize will be delayed in Period 1. The proof builds on expression (25) and follows the proof for Proposition 4 from the previous subsection. Given the possibility of aid, N- and T-agents, 19
respectively, will be indi¤erent if: U1 S1N+ReN+U 1 S1N+RAeN= Uf(1 )eN("1)(1 1)gg+U f[1 (S 2+R)AB]eNg Uf[(1 ) (S1TR)]eTg+U f[(1 ) (S1TR)A]eTg= U(1 )eT("1)1gc(h1; H1)+U f[(1 ) (S 2R)AB]eTg Linearizing and simplifying as in the previous section, yields the indi¤erence levels corresponding to (18) and (19) and the condition under which delay occurs. S1N=1 1 + A R+ 1 [1 + (1 B)] +("1) + (1 1)g eN +S 2(26) S1T=1 1 + A R+1+[1 + (1 B)] 1 ("1) + 1g+c(h 1; K) (1 )eT +S 2 (27) 2R > [1 + (1 B)] (1 )+("1) + (1 1)g eN +("1) + 1g+c(h 1; H) (1 )eT (28) To establish the proposition, compare the expressions (20) where B= 1 and (28), recall that < 1) < 1so that B < 1, which implies that, conditional on a recession, the right hand side of (28) is lesser in magnitude. Alternatively, if = 1 so that = 1, then the value of Breturns to 1 and the likelihood of delay is unchanged. In sum, this subsection has established that, while foreign aid conditional on a recession expands the set of exchange rate values that allow for an agreement in Period 2, it also increases the chances of a delay in stabilization in Period 1. Intuitively, the expectation that things will be bad enough in Period 2 so as to necessitate foreign aid increases the temptation to delay an agreement. Once delayed, however, it also makes an agreement more likely in Period 2 since aid alleviates some of the pain of adjustment. 3.3 Exchange Rate Expectations and Pre-Announced Aid Some countries have internal negotiations in conditions where it is already clear that the IMF will have to be approached in the near future for stabilization funds. Moreover, the broad contours of IMF conditionality, such as steps to change the …scal outlook and to address balance of payments problems with the help of exchange rate changes are more or less known due to past history with the institution and ongoing negotiations. 20
Consider a situation of foreign exchange scarcity that forces the government to strictly limit foreign currency access at the o¢ cial rate. A (possibly illegal) parallel open market exists where expectations of jump changes in the exchange rate can in‡uence its current level. To keep things simple, suppose that N-agents are forced by regulations to purchase foreign currency to pay for imports from this market while T-agents can purchase currency at the o¢ cial rate (which, at S= 1, is lower than the market rate in period 1). This could be either due to laws that allow privileged access to foreign currency for tradable producers –a not uncommon phenomenon –or because the T-agents can employ their foreign currency savings deposited in foreign banks (partly through misinvoicing). Suppose too that it is essentially certain that, unless there is stabilization in Period 1, the IMF will have to be approached in Period 2, and that the IMF will demand a devaluation to a level SIMF 2(which could correspond to the level that eliminates the parallel market premium) and consequent uni…- cation of the dual currency market. Thus, there is now no bargaining over the exchange rate between T- and N-agents in Period 2, and there is no uncertainty about the future level of the exchange rate. How will this a¤ect the parallel market exchange rate and the chances of an agreement in Period 1? We …rst need to determine what the market exchange rate in Period 1 will be given expectations of devaluation in Period 2. Some version of interest parity would be a good candidate but hard to justify rigorously in the absence of domestic and foreign assets …nancial assets in our framework. As an approximation, suppose that the level of the exchange rate in the parallel market is 1 + x, where xis the open market premium, and 1<1 + x<SIMF 2. Assume also, again plausibly, that this expected depreciation is a positive function of the exchange rate that the IMF is known to prefer for Period 2 if there is no stabilization in Period 1, i.e., x=x(SIMF 2),x0>0. To summarize, the set-up now is di¤erent from the previous two subsections in two respects: (1) the uncertainty about S2is now gone, and (2) the open market exchange rate in Period 1, S1, is now a function of the exchange rate level that the IMF is known to recommend as part of stabilization in Period 2. In order to isolate the e¤ects of the removal of exchange rate uncertainty, I set and from section 3.2 back to 1 (as in section 3.1). Proposition 8: Given ,x > 0, the presence of a known, conditionality-required level of the exchange rate SIMF 2that would remove exchange rate uncertainty in Period 2 has ambiguous e¤ects on the likelihood of delay. Proof: Since there is now no negotiating over, or uncertainty about, S2, the only interesting change is that Proposition 4 from Section 3.1 needs to be altered.9 The expressions that de…ne indi¤erence for each set of agents now become: 9That is, Propositions (1)-(3) are una¤ected and continue to hold. 21
(1 + )U1 S1N+ReN= U1 (1 + x(SIMF 2))eN("1)(1 1)g+U 1 SIMF 2eN (29) (1 + )Uf[(1 )(S1TR)]eTg= U(1 )eT("1)1gc(h 1; H1)+U (1 )SIMF 2eT(30) Again, linearizing and solving for the two exchange rates ( S1Nand S1T) yields, S1N=R+1 1 + (1 + x) +("1) + (1 1)g eN +SIMF 2(31) S1T=R+1 1 + 1 + 1 ("1) + 1g+c(h 1; K) eN +SIMF 2(32) and recognizing that a delay in stabilization requires that S1N< S1T, yields the following condition: 2R > 1 1 + xSIMF 2 (1 )+("1) + (1 1)g eN +("1) + 1g+c(h 1; K) (1 )eT (33) As long as the discount rate and the premium xare positive, the right hand side of the expression above may be lesser, equal to, or greater than that from (20). Comparing with the corresponding expressions in Section 3.1, i.e., (18) and (19), reveals that the removal of uncertainty about the exchange rate in Period 2 has ambiguous e¤ects on S1Nwhile increasing S1T. The overall impact on the likelihood of delay is ambiguous. First recall that the uncertainty about the level of the exchange rate should stabilization occur in Period 1 still exists. Intuitively, now there are two opposing e¤ects of removing uncertainty about S2and, therefore, causing movements today in the parallel market. The presence of an open market premium in Period 1 on the exchange rate makes it less costly for N-agents to agree to a depreciation in Period 1. It makes less sense to desire a delay if part of the costs of an agreement have already been borne in Period 1. On the other hand, the removal of exchange rate uncertainty in Period 2 and the resulting higher open market exchange rate prevalent in Period 1 reduces the potential cost of delaying.10 The net impact on 10 That is, instrument uncertainty now only has an e¤ect on the current period utility but not on the discounted second period utility. 22
the likelihood of delay will be the sum of these two forces. Perhaps surprisingly, the higher the exchange rate known to be required under the conditionalities of the program, the less likely a delay. Intuitively, if the parallel market rate available to N-agents already prices in the possibility of an agreement in Period 2, the cost of agreeing to stabilization today declines. 4 Concluding Remarks Large sustained …scal and balance of payments de…cits culminating in homegrown or IMF-in‡uenced stabilization programs are a common occurrence in developing countries. Often these programs seem to come at the last possible moment when the economy is on the verge of or in the middle of a crisis. This applies to …scal crises as well as exchange rate crises. Interestingly enough, this happens even though almost everyone can see the crises approaching, and some of the participants opposed to changes in exchange rate regimes oppose reforms in spite of the recognition on their part that delay could make their position even worse when the much needed reforms are eventually carried out. Why then the push to delay? This paper builds on existing literature by focusing on the political economy of changes in exchange rate levels and the resulting asymmetric e¤ects on di¤erent groups of producers. Producers of tradables and non-tradables have varying preferences over this key price in the macroeconomy. I show that, exchange rate uncertainty, even when di¤erent risk-averse groups have perfect foresight otherwise, can dominate the costs of in‡ation to an extent where it causes delays in stabilization. If in‡ationary costs are high, fear of the devaluation that follows stabilization increases the likelihood of delay and shrinks the space for agreement in the presence of a high share of imported inputs. IMF aid that reduces the burden of adjustment makes it more appealing to delay an agreement to the date when the aid is dispersed even if that happens following an economic downturn. An IMF conditionality that uni…es the o¢ cial and parallel exchange rate markets at a widely expected level and takes uncertainty out of the picture for the future exchange rate, on the other hand, could have ambiguous e¤ects on the likelihood of delay. That the devaluation is already partly priced into the current exchange rate increases the cost of delay for the group that does not have privileged access at the o¢ cial exchange rate. That the discounted cost of exchange rate uncertainty in the future is zero encourages delay. Thus, the higher the premium already incorporated into today’s parallel market exchange rate, the smaller the likelihood of delay. In sum, the form that foreign aid takes matters. It would be useful to end by noting some limitations of the analysis here. An important impact of external support is that it can in and of itself increase the credibility of a program. I only take this into account in a limited sense in section 3.3 by postulating that the IMF program removes exchange rate uncertainty in the second period . Also, IMF programs could cause moral hazard problems (Dreher and Vaubel (2004)) that are beyond the scope of this paper. 23