The new open economy macroeconomics: a survey

A common modelling framework – “the introduction of nominal rigidities and market imperfections into a dynamic general equilibrium model with well-specified microfoundations” – in which imperfect competition plays three distinct roles (§1). First, “in contrast to perfect competition (under which agents are price-takers), monopoly power permits the explicit analysis of pricing decisions.” Second, “equilibrium prices set above marginal cost rationalize demand-determined output in the short run, since firms are not losing money on the additional production.” Third, “monopoly power means that equilibrium production falls below the social optimum, which is a distortion that can potentially be corrected by activist monetary policy intervention” – with an immediate footnoted qualification that this “is only true if the shock is not so large as to drive marginal costs above marginal revenues.” The survey situates the programme alongside closed-economy developments labelled “neomonetarism” by Kimball and the “new neoclassical synthesis” by Goodfriend and King, and credits Svensson and van Wijnbergen (1989) as “[a]n important precursor” whose analytic structure was largely adopted, differing mainly in that it modelled outputs as stochastic endowments rather than endogenising production.

A two-country model of yeoman-farmers producing differentiated goods with one-period preset prices, whose results follow tightly from its strong symmetry assumptions (§2). Identical preferences and the law of one price mean “purchasing power parity holds and the consumption-based real exchange rate is constant”; agents trade a riskless real bond at a constant interest rate; there is no capital. For a Dornbusch-style unanticipated permanent home money increase: “[t]he impact effect of the monetary shock is an increase in the level of domestic output and consumption. The world real interest rate falls and nominal depreciation translates into a decline in the domestic terms of trade: both factors generate an increase in foreign consumption. The impact on foreign output is ambiguous, since the increase in aggregate consumption and the relative price shift work in opposite directions. The domestic current account moves into surplus.” Long-run non-neutrality follows from that surplus: it “implies a permanent improvement in domestic net foreign assets,” which finances a permanent trade deficit, and the wealth effect “reduces domestic labor supply (leisure is a normal good) and domestic output, thereby generating a permanent improvement in the home country’s terms of trade.”

Because purchasing power parity forces identical consumption growth across countries, which ties the short-run change in relative real balances to the long-run change (§2). Under PPP, “domestic and foreign agents face the same real interest rate and domestic and foreign consumption growth rates are thereby identical.” Writing the short-run and long-run monetary equilibrium conditions with that restriction, “it follows that, since the change in the money stock is permanent, the short-run change in relative domestic real balances must equal the long-run change and so the permanent increase in the nominal exchange rate just equals its initial jump.” This is the cleanest illustration of the survey’s general point: overshooting is absent not because the model denies it but because PPP has been assumed, and it reappears as soon as national price levels are delinked.

Because the first-order effect is a general rise in world demand against a pre-existing monopolistic distortion, while all the asymmetric channels are second-order (§2). “Remarkably, it turns out that home and foreign welfare are raised by the same amount, despite the asymmetric output effects of the shock. The intuition for this result is that the first-order effect of the monetary shock is the initial general increase in world demand. Since the imperfect competition distortion means that the initial level of output was too low, a demand-driven increase in world output raises welfare, to the equal benefit of both countries.” The asymmetries vanish by an envelope argument: “the fact that home agents produce more does not raise their relative utility level: the extra revenue is exactly cancelled out by the increase in work effort,” and “current-account imbalances have only second-order effects, since the initial equilibrium leaves unexploited any marginal gains from reallocating consumption and leisure across time periods.” The survey draws the methodological moral: “[t]he surprising result that both countries gain equally from an unexpected domestic monetary expansion illustrates the utility-based evaluation offers a non-trivial advantage over traditional ad-hoc loss functions.”

Exogenously and time-dependently, with the survey flagging both the arbitrariness and the reason for it (§3). “The literature typically introduces nominal rigidity as an exogenous feature of the environment. In the Redux model, firms simultaneously set prices one period in advance. This assumption is arbitrary but convenient, since all adjustment is completed after just one period.” Two limits are noted. If stickiness rests on a fixed menu cost, “firms will be motivated to immediately adjust prices in the event of a large enough shock,” and Corsetti and Pesenti emphasise that a sufficiently large shock “would violate firms’ participation cost by raising marginal cost above price,” so “the analysis should be interpreted as applying only to the relevant range of shocks.” And the form of rigidity is dictated by tractability: “nominal rigidity is invariably modelled in this literature as of the time-dependent variety, since state-dependent pricing is not easily incorporated into general equilibrium models.” On sticky wages instead of prices, Hau’s case in which prices are flexible but nominal wages predetermined turns out to be observationally similar, since monopolistic firms set prices as a constant markup over the wage so “optimal prices also remain fixed in the short run.”

Not by itself – the survey is explicit that this is a common misreading (§3.2). Calvo pricing gives a smooth price level (“if the Poisson arrival rate of a price-change opportunity is gamma, a fraction gamma of firms changes its price each period and 1/gamma is the average interval between price changes”) and is “a potential persistence mechanism since the adjustment to a shock cannot be achieved instantaneously.” But “Chari et al. (1998a) show that staggering in itself does not generate endogenous persistence if prices are a constant markup over marginal costs and if marginal costs are increasing in the level of output. Under these conditions, a firm will raise its price as soon as it is given the opportunity.” The general principle the survey extracts is that “the responsiveness of prices and persistence depend on (i) the sensitivity of prices to costs and (ii) the sensitivity of costs to output.” Hence the mechanisms that do work: convex demand schedules, so that “the price elasticity of demand is increasing in the price charged” and firms are slower to raise prices; rigid real wages, since “if marginal costs are rigid, optimal prices will also be sticky” (Jeanne); and wage rather than price staggering, which Andersen argues is more likely to generate persistence because “wage stickiness implies that labor demand rather than labor supply determines quantities in the labor market,” making the labour supply elasticity irrelevant for short-run marginal costs.

The currency of price stickiness, not segmentation alone (§4, §4.1). Segmentation means some firms can charge different prices in the two markets, but “[w]ith identical CES preferences across countries, even these firms will optimally select home and foreign currency prices that are a constant markup over marginal cost and hence the law of one price will be satisfied ex ante.” The second ingredient is essential: “prices that are sticky in each local currency means that exchange rate movements cause ex-post deviations from the law of one price.” With a fraction s of firms pricing to market (Betts and Devereux), the expenditure-switching effect of exchange rate movements disappears, so “the size of the exchange rate movement required to satisfy the monetary equilibrium condition is enlarged. This raises the possibility of short-run exchange rate overshooting, which is ruled out in the Redux model.” The comovement implications improve the fit to international business-cycle data: real exchange rate movements delink home and foreign consumption growth while “the correlation of home and foreign output rises since the domestic demand expansion raises demand for imports at the fixed relative price of imports.” Two sharp reversals follow: with full pricing to market “the current account remains in balance, contrary to the surplus prediction in the Redux model,” and “an exchange rate depreciation can actually improve a country’s terms of trade,” so the expansion becomes beggar-thy-neighbour.

Because they break the constant markup, which is what allows both real exchange rate and law-of-one-price deviations to persist after every firm has repriced (§4.2). Bergin and Feenstra depart from constant-elasticity demand: under translog preferences “the expenditure share for each good j is inversely related to its relative price, which generate variable markups over marginal cost,” and following Basu they add intermediate goods “so that marginal costs are heavily influenced by the aggregate price level.” The result is that “monetary shocks have persistent effects on real exchange rates, even after all firms have had the opportunity to adjust prices,” because each firm is doubly reluctant to raise its price – doing so cuts its expenditure share, and other firms’ fixed prices keep intermediate costs from rising. The survey draws the contrast explicitly: “[t]his stands in contrast to the other PTM models that specify a constant elasticity of demand (and thereby a constant markup): in that case, once firms are free to adjust prices, the law of one price will be re-established. As such, PTM does not in itself generate endogenous persistence beyond the length of the exogenous nominal rigidity in the model.” A further subtlety is that persistence and volatility interact in both directions: persistent interest differentials require a more extreme initial exchange rate jump, but “a larger exchange rate response alters marginal costs and hence induces faster price adjustment, reducing the persistency of the impact of a monetary shock on real variables.”

A great deal, along several dimensions the survey separates (§5). The Redux assumption that all varieties enter a single symmetric CES index is unusual; Chari et al. instead set the elasticity of substitution between home and foreign goods below that between varieties within a category, with marked home bias. Corsetti and Pesenti’s unitary home-foreign elasticity is analytically pivotal: constant income shares mean “[t]he risk-sharing provided by such offsetting terms of trade movements renders securities markets redundant and the equilibrium current account is always zero,” which “allows the authors to write a version of the Redux model that is solvable in closed form, without resorting to linearizations,” and hence to analyse non-infinitesimal shocks and derive optimal policies as functions of structural parameters. The consequence for policy is substantive: “a domestic monetary expansion can be ’too large’ in the sense that the positive welfare effect of an increase in home output can be more than offset by a decline in its terms of trade,” so the optimal monetary surprise is finite and a country will not expand to competitive output. On the consumption elasticity of money demand: irrelevant to exchange rate volatility under PPP, it “reemerges as a key parameter” under pricing to market, with volatility an inverse function of it, by the Dornbusch logic that a low elasticity requires the interest rate to fall, which requires expected appreciation. On consumption-leisure separability: non-separable preferences consistent with balanced growth imply a monetary shock raises relative domestic consumption to compensate for extra work effort, “mitigating the impact of monetary shocks on the real exchange rate,” so Chari et al. “argue that preferences separable in consumption and leisure are required in order to explain high exchange rate volatility,” with balanced growth then requiring that non-market productivity rise in line with market productivity and that the labour-supply and intertemporal elasticity parameters coincide.

Each overturns a specific Redux result (§5.4, §5.5). Capital: “monetary shocks can cause investment booms by reducing the short-run interest rate. In turn, this means that a monetary shock may actually generate a current account deficit, in contrast to the Redux prediction of a surplus,” while raising current relative to future labour supply so that “the persistence of shocks is diminished.” Non-traded goods (Hau): the initial exchange rate response is larger, “since non-traded prices are tied down by the sticky nominal wage, [so] a larger exchange rate movement is required to accomplish a given change in the aggregate price level,” though two offsetting channels (home-biased demand expansion, and a low domestic consumption-based real interest rate as the real exchange rate is expected to appreciate) “partly offset the increase in exchange rate volatility.” The survey adds a falsification point against this mechanism: “high exchange rate volatility in Hau’s model is coterminous with high volatility of the relative price of imports relative to domestically produced goods, which is not a pattern clearly observed in the data.” Home bias in tradables (Warnock): “a domestic monetary expansion improves home welfare by more than foreign welfare, so that the gains are not equally shared,” because net foreign asset accumulation now has wealth effects on relative prices through demand composition as well as through labour supply, making the real exchange rate non-constant and permitting overshooting.

Less than one might expect for monetary shocks, for a quantitative reason – but qualitatively, and for fiscal shocks, it matters (§6). Complete markets simplify by removing the propagation channel: “full risk-sharing means that there are no shifts in wealth arising from monetary shocks,” eliminating “the current account and net foreign assets as a dynamic propagation mechanism.” The survey records Obstfeld and Rogoff’s objection to assuming completeness – “it would be strange to analyze imperfections and rigidities in goods markets but at the same time assume the completeness of international capital markets” – and its converse, that “it is hard to imagine how price or wage rigidities could survive in a world sufficiently sophisticated that complete international risk sharing is accomplished.” Empirically within the models, “the incompleteness of financial markets makes little difference for the persistence of monetary shocks in this context, since equilibrium current account movements are small,” and Betts and Devereux find the same even when the law of one price holds – though a footnote records that “the asset market structure is extremely important for the transmission of fiscal shocks,” where by contrast the denomination of sticky prices matters little. Tille shows financial structure does qualitatively alter the current account response under pricing to market. Trading frictions (Sutherland) let the domestic interest rate deviate from the foreign one, producing exchange rate undershooting and “lower volatility in the exchange rate and consumption but larger volatility in output and interest rates,” with an interaction worth noting: “barriers to financial integration have a larger impact, the greater the degree of price inertia,” because slower output adjustment makes agents want more consumption smoothing. Senay finds the two integrations interact: “financial market integration reduces the output response to a monetary shock if goods markets are segmented but actually increases the output response if goods markets are integrated.”

That the sign of spillovers, and hence the direction of the gains from coordination, depends on the pricing assumption – and that even within one model the policy reaction function’s sign depends on two elasticities (§7). The framing is Cooper and John’s: are spillovers positive, and are policies strategic complements or substitutes? “In the Redux model, monetary policies have positive spillover effects but home and foreign monetary policies are strategically independent.” In Corsetti-Pesenti spillovers stay positive, but “the sign of the policy response function depends on the interplay between the intertemporal elasticity of substitution and the intratemporal elasticity of substitution between home and foreign goods”: if the intertemporal elasticity is larger, policies are strategic substitutes and the home bank contracts in response to a foreign expansion, because the foreign expansion “raises domestic output relative to its initial optimized value, imposing an excessive cost in terms of foregone home leisure”; if smaller, the home bank expands; if the two coincide, “the best response to a foreign policy shock is to ‘do nothing’.” Efficient output requires coordination, since “a country undertaking a unilateral expansion would take into account the negative terms of trade effect and hence choose a tighter policy”; Benigno shows non-cooperation therefore “imparts a contractionary bias,” and that under a supranational authority constrained to Pareto-improve on Nash, “the larger country is pushed to the competitive output level while the smaller country may retain the monopolistic distortion.” Under full pricing to market the sign flips: “monetary policy actually exerts a negative spillover effect,” so “with full pricing to market, policy spillovers are negative and policy coordination leads to a slower rate of monetary expansion,” which “restores the rationale for international policy coordination, in the sense that it leads to a costless decline in the common world inflation rate” – reversing Rogoff’s (1985) argument. The survey also notes Devereux’s point that pricing to market undercuts the standard case for floating: “the merits of exchange rate flexibility as an adjustment mechanism are sharply reduced by PTM, since exchange rates do not alter relative prices,” so “a country that fixes its exchange rate does not necessarily suffer increased output volatility.”

First-order welfare effects of monetary volatility through price setting, the first utility-based comparison of exchange rate regimes, and by-products for asset pricing (§8). With log-normal money processes in the Corsetti-Pesenti structure, Obstfeld and Rogoff show that “if the home country faces monetary uncertainty, its firms incorporate a risk premium into output prices, depressing production but improving the terms of trade. In this way, uncertainty has first-order effects on ex ante welfare levels.” A notable symmetry result follows – “home and foreign countries have the same incentives in designing an optimal global exchange rate system” – and, the survey stresses, “this result holds even if the home and foreign countries are different in terms of relative size, contrary to the usual presumption that small countries should care more about exchange rate stability.” On asset pricing, “the risk premium on a volatile currency may actually be negative if exchange rate movements hedge consumption volatility,” yielding “a novel explanation of the forward premium puzzle.” Devereux and Engel extend this to pricing to market and compare regimes on welfare rather than on an ad hoc loss function: “[s]ince PTM insulates consumption from exchange rate fluctuations, floating exchange rates are less costly under PTM than under producer currency pricing and a flexible regime will dominate pegging whenever agents are at least as risk averse as logarithmic consumers (which is the empirically relevant range). Under producer currency pricing, in contrast, fixed exchange rates will be preferred if risk aversion is sufficiently high.” Bacchetta and van Wincoop formalise the old claim that exchange rate uncertainty reduces trade, showing that when consumption and leisure are substitutes firms charge a higher price abroad, so “production is orientated to the home market and the volume of international trade declines” – and that this survives complete financial markets, since forward hedging “guarantees the domestic currency value of a given amount of foreign currency revenues, [but] cannot undo the impact of uncertainty about the level of any such foreign currency revenues.”

A version with monopolistic competition and sticky prices only in non-tradables, which cleanly separates mechanisms the two-country models entangle (§10). In the Obstfeld-Rogoff appendix model, traded output is an endowment whose domestic price is the world price times the exchange rate, tradables and non-tradables enter log-separably, and the discount rate equals the world interest rate – so “the optimal path for tradables consumption is perfectly flat, so that the current account always remains in balance” and money is neutral in the long run. Overshooting nonetheless occurs, and the survey works through why: the condition turns out to be “precisely the same overshooting condition as in the original Dornbusch (1976) model,” namely a consumption elasticity of money demand below one. Replacing log-separability with a CES aggregate over traded and non-traded goods (Lane 1998) restores current account movement, with the sign determined by two parameters: “[i]f [the intertemporal elasticity] < [the intratemporal elasticity], the latter effect dominates: the rise in non-traded output and consumption leads to a decline in traded consumption and a positive monetary shock thereby generates a current account surplus. In contrast, a current account deficit occurs if [intertemporal] > [intratemporal]. Finally, the current account remains in balance only if [they are equal].” The survey’s gloss is its own thesis in miniature: “the model illustrates that traditional Mundell-Fleming results are sensitive to the precise specification of a model’s microfoundations.”

Some calibration, some VAR evidence, and a scattering of indirect tests – with the survey sceptical that moment-matching can settle the question (§11). On matching unconditional moments, the survey raises two objections: “even if monetary shocks only account for a fraction of the aggregate economic fluctuations over a given time interval, this is not inconsistent with the existence of nominal rigidities or an important role for monetary policy in responding to other disturbances,” and “it is widely accepted that the unconditional variances of nominal and real exchange rates are infected by considerable market noise that is unrelated to macroeconomic fundamentals” – so “this calibration method is not sufficient in obtaining an overall empirical evaluation of this class of models.” VAR impulse responses fare better: Clarida and Gali and Eichenbaum and Evans show monetary shocks move the real exchange rate “in a manner that is qualitatively consistent with the predictions of sticky-price models,” and a calibrated pricing-to-market model “clearly outperforms the PPP-based Redux model.” On current account responses, the survey is careful about identification: Lane imposes that monetary shocks have no long-run effect on the current account, while Prasad and Lee-Chinn assume no long-run effect on the real exchange rate, “a condition that holds only in a subset of the models reviewed above.” Broadly, “each of these papers find that positive nominal shocks tend to improve the current account position,” with a J-curve response of the trade balance. Indirect tests include the openness-inflation relation, Hau’s finding that “real exchange rate volatility and openness are indeed inversely correlated in a large cross-section of countries, even when openness is treated as endogenous,” and evidence of a significant net-foreign-assets/real-exchange-rate relation among OECD economies, which “provides indirect support for the notion that even temporary disturbances can have persistent effects.” Two gaps are named: parameter values, on which “Summers (1991) is sceptical that such parameters can be estimated with any precision,” and the pricing assumption itself – “[w]e know surprisingly little about the extent of PTM behavior in the data.”

That the framework’s advance is real but that its policy usefulness is not yet established, and that this is a problem for the programme rather than a caveat (§12). The advances: dynamic tracking of effects rather than impact effects only; “the solid microfoundations embedded in these models permit welfare analysis, which can generate some surprising results”; and stochastic versions “well-designed for making meaningful comparisons across different policy regimes.” The warning is stated at full strength: “As is readily apparent from this survey, many welfare results are highly sensitive to the precise denomination of price stickiness, the specification of preferences and financial market structure. For this reason, any policy recommendations emanating from this literature must be highly qualified. This is an issue of some concern, since the new open economy macroeconomics will be of only limited interest in policy circles unless researchers converge on a ‘preferred’ specification that is buttressed by extensive supporting empirical evidence.” Note what this does and does not claim: it is a statement about the state of the literature in 1999-2000, not a claim that the models are wrong, and the survey ends by expecting the field to grow.