Fiscal Cyclicality and Currency Risk Premia

Abstract

I develop a model of real exchange rate determination that attributes a central role to the intertemporal government budget condition, which equates the market value of government debt to the present value of government surpluses. To enforce this equilibrium condition in the presence of nominal rigidities, the real exchange rate has to adjust in response to shocks to government surpluses. The model predicts that fiscal shocks account for real exchange rate movements, and the factor structure in fiscal shocks aligns with the factor structure in currency returns. Both predictions are confirmed in the sample of developed countries.

Textbook open-economy models imply a strong link between government fiscal conditions and real exchange rates. Under standard assumptions, a higher government spending crowds out private consumption and leads to real appreciation of the local currency. As Monacelli and Perotti (2010) show, this relationship between government spending and real exchange rate is a robust feature of the textbook models, regardless of the presence of investment and of the assumed degree of price stickiness. Empirically, however, identified government spending shocks are associated with real depreciation of the local currency.

In this paper, I adopt an asset pricing approach to study the relationship between fiscal conditions and exchange rates. I demonstrate my results in a tractable two-country model in which governments issue nominal debt and prices are sticky. The key ingredient is the intertemporal government budget condition, which equates the real value of a government’s debt to the present value of its current and future primary surpluses. This equilibrium condition is a present value formula, which relates the market value of an asset—the government debt—to the present value of its cash flows—the government surpluses.

The present value of government surpluses is the sum of expected government surpluses divided by the appropriate discount rates. Under the standard asset pricing logic, the asset value adjusts in response to changes in cash flows and discount rates. In this context, however, nominal rigidities constrain the extent to which the real value of government debt can adjust. As a result, when the cash flows—that is, the government surpluses—increase, the real discount rates have to increase in order to satisfy the intertemporal government budget condition. This increase in the real discount rates triggers real appreciation of the local currency, giving rise to a fiscal mechanism of real exchange rate determination.

This novel mechanism predicts that increases in both current and expected future government surpluses are correlated with local currency appreciation. Since the government surplus is tax revenue minus government spending, the relationship between current government surpluses and real exchange rate movements is consistent with the empirical literature that finds real exchange rate depreciation occurs following a positive government spending shock.

The relationship between expected future government surpluses and real exchange rate movements is tested in this paper. I obtain the International Monetary Fund’s (IMF) government surplus forecasts as a proxy for the expected future government surpluses. Across 11 developed countries from 2009 to 2018, an increase in the government surplus forecast normalized by the government debt quantity is correlated with both real and nominal exchange rate appreciation. In a horse race with other fundamental variables, including gross domestic product (GDP) forecasts, consumption growth, money supply, current account, and currency forward discount, the government surplus forecasts remain the only variable that explains real exchange rate movements. Also, in a longer sample from 1980 in which the IMF forecasts are not available, I confirm this present value relationship using alternative approaches.

Moreover, I find that different countries’ government surpluses load on a common factor with different loadings. Consistent with my model’s prediction, this factor structure in government surpluses also aligns with the factor structure in currency returns. More precisely, I use the cross-sectional average of the changes in government surpluses as a proxy for the common factor in government surpluses, and use the currency carry trade return as a proxy for the common factor in currency returns. I find that a country whose government surplus is more exposed to the common surplus factor also tends to have a currency that is more exposed to the carry trade return. The governments’ fiscal exposures account for 44% of the variation in carry trade betas in the post-2009 data of government surplus forecasts, and for 58% of the variation in the post-1980 data of realized government surpluses. These results further support the model’s prediction that exchange rates comove with government surpluses.

In summary, this paper relates government fiscal conditions to real exchange rate movements, both theoretically and empirically. It illustrates how the intertemporal government budget condition governs the joint dynamics of government surpluses and exchange rates, thereby providing an explanation of exchange rate movements and exchange rate factor structure based on fiscal fundamentals.

For tractability, I make simplifying assumptions in my model, such as log preferences and complete markets. These assumptions generate perfect correlations between the government surplus, the real exchange rate, and consumption. On the one hand, my model predicts that a higher government surplus is associated with a lower concurrent consumption. This prediction is consistent with the empirical finding that a positive spending shock raises private consumption, whereas textbook models fail to generate not only the aforementioned spending-exchange rate correlation but also this spending-consumption correlation (Monacelli and Perotti 2010).

On the other hand, like textbook models with perfect risk sharing, my model predicts that a country’s real exchange rate should depreciate when its consumption grows relative to foreign countries, and the correlation is perfect. This prediction runs head-on into the Backus and Smith (1993) puzzle, the observation that this correlation is close to zero in the data. Prior resolutions of the puzzle include demand shocks (Pavlova and Rigobon 2007) and market segmentation (Alvarez, Atkeson, and Kehoe 2002, 2009). In this paper’s context, one potentially interesting resolution is to adopt a richer structure of household preferences that allows for endogenous and time-varying risk premia. When the discount rates on the government surpluses need to adjust to enforce the intertemporal government budget condition, the adjustments can happen in the risk premia instead of the concurrent consumption. In this way, fiscal shocks can drive real exchange rate movements via the risk premium channel, without the need to generate counterfactual predictions for consumption movements. I leave this extension for future work.

A large literature studies the effect of fiscal policy on real and nominal exchange rates. For example, Ravn, Schmitt-Grohé, and Uribe (2007), Kim and Roubini (2008), Monacelli and Perotti (2010) study how identified government spending shocks affect exchange rates. Daniel (2001a), Burnside, Eichenbaum, and Rebelo (2001, 2003); Corsetti and Mackowiak (2001) study the fiscal root of currency crisis. Koijen and Yogo (2020) adopt a demand system approach to attribute exchange rate variations to macro variables, including fiscal conditions. While a large class of standard open-economy models since Mundell-Fleming predicts that a country’s real exchange rate should appreciate following a positive government spending shock, evidence suggests the opposite. As Monacelli and Perotti (2010) point out, this puzzle poses a theoretical challenge since models with incomplete markets, nonseparable utility, and variable markups have made only limited progress.

My model provides a resolution to this puzzle by developing a new mechanism of exchange rate determination that imputes a central role to the intertemporal government budget condition. As discussed in the introduction, this mechanism generates a comovement between increases in government surpluses and real appreciation of local currencies, which is consistent with the empirical findings in this paper and in the previous literature.

In doing so, my paper also helps us understand why exchange rates are disconnected from economic fundamentals (Meese and Rogoff 1983; Backus and Smith 1993; Obstfeld and Rogoff 2000). In my sticky-price model, equilibrium exchange rates load on fiscal shocks, and do not load on productivity shocks. Previous attempts to resolve this puzzle include models of investor preferences, market structure, and monetary and financial frictions. For example, Colacito and Croce (2011, 2013), Bansal and Shaliastovich (2013), Colacito et al. (2018) feature long-run risks; Verdelhan (2010), Heyerdahl-Larsen (2014), and Stathopoulos (2017) feature habits; Gourio, Siemer, and Verdelhan (2013); Farhi and Gabaix (2016) feature rare disasters; Gabaix and Maggiori (2015), Itskhoki and Mukhin (2017), and Jiang (2019) feature financial frictions; Chien et al. (2015) and Dou and Verdelhan (2015) feature market segmentation; and Jiang et al. (2018a, 2018b) feature demand for safe assets. Incorporating some of these elements into my model can help maintain the same insight, while making the model more realistic along other aspects.

Moreover, my paper documents a factor structure in government surpluses and shows it aligns with the factor structure in currency returns. Previous literature documents a factor structure in currency returns (Lustig Roussanov, and Verdelhan 2011; Hassan and Mano 2019; Verdelhan 2018), and regards the heterogeneous loadings on common factors as the key determinant for currency risk premia (Gourio, Siemer, and Verdelhan 2013; Colacito et al. 2018). Subsequent papers relate currency risk exposures to commodity exports (Powers 2015; Ready, Roussanov, and Ward, 2017a, 2017b), country size (Hassan 2013, Martin 2011), trade network (Richmond 2019, Jiang and Richmond 2019), and external debt position (Corte, Riddiough, and Sarno 2016; Wiriadinata 2018). My paper ties the heterogeneity in currency risk exposures to fiscal shocks.

A related literature presents alternative mechanisms through which taxes affect the expected level and the uncertainty of future growth rates, as well as marginal utilities via recursive preferences (Croce et al. 2012a; Croce, Nguyen, and Schmid; 2012b, 2013; Croce et al. 2019). My paper provides a different but complementary mechanism that connects government surpluses to real exchange rates via the intertemporal government budget condition.

Lastly, my model extends the fiscal theory of the price level in an international context. This literature mainly focuses on how the present value of government surpluses is related to domestic price levels (Sargent and Wallace 1984; Leeper 1991; Woodford 1994; Sims 1994; Cochrane, 2005, 2017, 2018; Corhay et al. 2018). While there have been efforts to extend this theory to the open economy (Dupor 2000; Daniel 2001b), my paper lays out a novel mechanism that relates government fiscal conditions to real exchange rates. In particular, the fiscal theory of the price level claims that the price level responds one-for-one to the present value of government surpluses. My model shows that adding sticky prices turns this result on its head: instead of the price level, consumption and real discount rates adjust in response to fiscal shocks. In this sense, my model is a fiscal theory of real adjustments.

1. Model

In this section, I present a two-country model with nominal rigidities, which follows Corsetti and Pesenti (2005) until I introduce the intertemporal government budget condition. I will first characterize the model under flexible prices that generates the standard results representative of those obtained in textbook models. Then, I will characterize the model under sticky prices and explain its novel implications.

1.1 Environment

1.1.1 Household preferences

There are two countries, Home and Foreign. Each country contains a unit mass of households, a unit mass of firms, and a government. Home households are indexed by |$j\in [0,1]$|⁠, and Home firms are indexed by |$h\in [0,1]$|⁠. Foreign households are indexed by |$j^*\in [0,1]$|⁠, and Foreign firms are indexed by |$f\in [0,1]$|⁠. Each firm produces a unique variety of good, which is an imperfect substitute for other varieties.

The parameter |$\alpha>1/2$| measures home bias in consumption, and the parameter |$\theta$| is the elasticity of substitution across varieties. Foreign households face a similar problem, which is presented in Internet Appendix A.1.

Let |$p_t(h)$| and |$p_t(f)$| denote the Home-currency prices of varieties |$h$| and |$f$|⁠. Let |$P_{H,t}$| and |$P_{F,t}$| denote the Home-currency prices of the Home and Foreign bundles |$c_{H,t}(j)$| and |$c_{F,t}(j)$|⁠. Let |$P_t$| denote the Home-currency price of the aggregate consumption basket in the Home country. These prices are derived in Internet Appendix A.2.

1.1.2 Household budget constraint

Home household |$j$| owns the portfolio of Home firms and provides labor to the firms. It earns a nominal wage |$W_t$| and receives a nominal dividend |$D_t(j)$| from the firms. It also pays tax |$\tau_t(j)$| to the government and purchases consumption |$c_t(j)$|⁠, both denoted in real terms.

The financial markets are complete. Let |$\sigma_t$| denote the state of the economy at time |$t$|⁠. Let |$Q(\sigma_{t+1}|\sigma_t)$| denote the time-|$t$| Home-currency price for one unit of Home currency delivered at time |$t + 1$| contingent on the state being |$\sigma_{t+1}$|⁠. At time |$t$|⁠, Home household |$j$| holds |$B_t(\sigma_{t+1},j)$| unit of the Arrow-Debreu security that pays off in state |$\sigma_{t+1}$|⁠. |$Q^*(\sigma_{t+1}|\sigma_t)$| is similarly defined as the time-|$t$| Foreign-currency price for one unit of Foreign currency delivered at time |$t + 1$| contingent on the state being |$\sigma_{t+1}$|⁠, and |$B_t^*(\sigma_{t+1},j)$| is the quantity of this security held by Home household |$j$|⁠.

The Home household’s Lagrangian is derived in Internet Appendix A.3. By symmetry, all households in each country have the same consumption, saving, and labor decisions. Therefore, I can drop their indices |$j$| and |$j^*$|⁠.

1.1.3 Firms

I consider a simple form of nominal rigidities: First, firms have to set prices one period in advance. Second, I assume Producer Currency Pricing, the case in which exports are priced and invoiced in the firms’ domestic currencies. The optimal price setting is derived in Internet Appendix A.4. The main result also holds if I instead assume Local Currency Pricing, the case in which exports are priced and invoiced in the consumers’ domestic currencies (see Internet Appendix A.11).

1.1.4 Governments

I make the following assumptions about the Home and Foreign governments. First, tax and spending fall on the same basket of goods as the households’ consumption, which is an aggregate of both Home and Foreign varieties. For the Home government, let |$\tau_t$| denote its tax revenue and let |$g_t$| denote its spending in real terms.

Second, government spending is not remitted back to domestic households. As will be shown in Section 1.2, under this assumption and flexible prices, government spending reduces the available goods for consumption and hence behaves like a negative supply shock. This standard mechanism will be contrasted with the new effects generated by the intertemporal government budget condition and sticky prices.

1.1.5 Market-clearing conditions

Note that |$B_{t+1}$| denotes the quantity of the Home government debt that is due at time |$t+1$|⁠. The market-clearing condition (3) requires that the total amount of Home currency that both countries’ households receive from their holdings of Arrow-Debreu securities in state |$\sigma_{t+1}$| is equal to the amount of nominal debt the Home government pays back at time |$t+1$|⁠. Internet Appendix A.1 presents the Foreign counterparts.

1.2 Characterizations under flexible prices

In this standard setup, international risk sharing allows Home households to benefit from both Home and Foreign productivity improvements. Because of the home bias in consumption (i.e., |$\alpha>1-\alpha$|⁠), the Home households’ consumption is more exposed to the Home productivity shock than to the Foreign one. Moreover, since the labor curvature coefficient |$\nu$| is positive, a higher labor effort drives up the real wage. As a result, a higher government spending makes labor more costly, and therefore crowds out household consumption.

1.3 The economics of the government budget condition under sticky prices

Next, I introduce sticky prices and show how the model generates the opposite result that a higher Home government surplus is associated with Home currency appreciation. For tractability, I make an additional assumption.

  

Internet Appendix A.8 provides the proof. Lemma 1 shows that the present value of future government surpluses |$\mathbb{E}_t\left[ \sum_{k=t+1}^\infty m_{t,k} s_k \right]$| is a constant |$A$| multiplied by the current consumption |$c_t$|⁠. Furthermore, while the government surplus can be negative ex post, I assume that the present value of future government surpluses is positive:

 

This equation contains the key insight of this model. The left-hand side represents the real value of government debt. |$B_t$| denotes the quantity of nominal debt issued at time |$t-1$|⁠, which cannot change at time |$t$|⁠. |$P_t$| is the aggregate Home price level. When prices are flexible, the price level |$P_t$| can freely adjust in response to fiscal shocks. When prices are sticky, the real value of debt |$B_t/P_t$| has sluggish adjustments. In fact, if the home bias in consumption approaches 1 (i.e., |$\alpha \rightarrow 1$|⁠), |$B_t/P_t$| cannot change at time |$t$|⁠.

The sluggish adjustments on the left-hand side of Equation (4) constrain the variation on the right-hand side, which represents the present value of current and future Home government surpluses. In particular, when the current Home government surplus |$s_t$| rises, the present value of future government surpluses has to decline in order to satisfy this intertemporal budget condition. Since future government surpluses follow an i.i.d. distribution, for the decline in the present value to happen, the discount rates on future government surpluses have to rise endogenously. In this way, under sticky prices, the intertemporal government budget condition (4) requires the Home discount rate to rise in response to an increase in the Home government surplus. Next, I show the increase in the Home discount rate leads to real appreciation of the Home currency.

1.4 Characterizations under sticky prices

I can further characterize the equilibrium dynamics under a log-linear approximation. I define the steady-state government surplus |$\bar s$| and the steady-state productivity |$\bar z$| as the means of their i.i.d. distributions. I define the steady-state consumption |$\bar c$| as the consumption obtained when |$s_t=s^*_t=\bar s$| and |$z_t=z^*_t=\bar z$|⁠. By symmetry, the steady-state Home and Foreign consumptions are the same.

 

Internet Appendix A.9 presents the proof and the expressions for the constants (⁠|$\kappa$|⁠, |$\kappa^*$|⁠, and |$\kappa^P$|⁠). Consistent with the intuition discussed in Section 1.3, Proposition 1 shows that a higher Home government surplus leads to a higher Home discount rate and lower concurrent Home consumption. Moreover, if there is home bias in consumption (⁠|$\alpha>1/2$|⁠), the Home currency appreciates (i.e., lower |$e_t$|⁠) when the Home government surplus, |$s_t$|⁠, increases and depreciates when the Foreign government surplus, |$s^*_t$|⁠, increases.

Because the state variables are i.i.d. and risk premia are constant, the expected currency excess return |$\mathbb{E}_t[rx_{t+1}]$| and the expected real exchange rate |$\mathbb{E}_t[\log e_{t+1}]$| are constant. So, the real exchange rate is determined by the real interest rate differential |$(r^*_t - r_t)$| in this simple model. A higher Home government surplus |$s_t$| increases the Home real interest rate |$r_t$|⁠, and therefore leads to Home real exchange rate appreciation (i.e., lower |$e_t$|⁠). Similarly, a higher Foreign government surplus increases the Foreign real interest rate, and leads to Foreign real exchange rate appreciation.

In comparison, the fiscal shock has a weaker effect on the price level. In fact, if the home bias in consumption approaches 1 (i.e., |$\alpha \rightarrow 1$|⁠), then the Home price level |$\log P_t \approx \kappa^P + \log B_t$| is not affected by the Home government surplus |$s_t$| at all. In this case, the real and the nominal exchange rates react to the shocks to government surpluses in the same way.

1.4.1 Discussions

What distinguishes this model from the textbook models is the combination of sticky prices and the intertemporal government budget condition. Under this specification, the real discount rate has to adjust in response to fiscal shocks in order to enforce the government budget condition. Below, I will contrast this mechanism with two traditional approaches.

(1) Flexible prices: As discussed in Monacelli and Perotti (2010), a robust feature of open-economy models since Mundell-Fleming is that a country’s real exchange rate appreciates following an increase in its government spending (or equivalently a decline in the government surplus). My flexible-price model in Section 1.2 illustrates the standard mechanism that generates this feature: an increase in government spending leads to lower private consumption, which, through international risk-sharing condition, implies that the real exchange rate must appreciate.

(2) Sticky prices, trivial government budget condition: In standard models with sticky prices (Corsetti and Pesenti 2005), if government spending is entirely financed by tax, the intertemporal government budget condition (4) holds trivially (government debt |$=$| government primary surplus |$=$| 0). The real discount rate does not need to adjust to enforce this condition. In this case, monetary shocks have first-order effects on the equilibrium consumption and exchange rates, whereas the fiscal mechanism that gives rise to my new results is muted.

Proposition 1 considers an isolated shock to today’s government surplus. It is possible that the government budget balances over long periods of time, so that a decline in today’s government surplus is offset by increases in future government surpluses. In this case, the present value of government surpluses remains constant without requiring any equilibrium adjustment in the discount rate. However, Jiang et al. (2020, 2021) examine the temporal responses of government surpluses following a fiscal shock or a GDP shock in the United States, and do not find evidence for the offsetting pattern.

1.5 Shocks to expected future government surpluses

For tractability, the following corollary considers an unanticipated shock that affects all future government surpluses while preserving the shape of their distribution, and shows how this shock affects the real exchange rate.

 

Internet Appendix A.10 gives the proof. When |$\delta>1$|⁠, the new distribution |$\tilde F$| spreads out the original distribution, raising the average Home government surplus in the future. This surprise increase in the expected future Home government surpluses (i.e., higher |$\log\delta$|⁠) then leads to real appreciation of the Home exchange rate (i.e., lower |$e_t$|⁠).

A novel aspect of this result is that the current government surplus and the expectation of future government surpluses both matter for the real exchange rate, which therefore behaves like a forward-looking asset price. The following thought experiment helps us understand the magnitudes of these effects. First, consider a uniform 1% increase in the Home government surplus in every single period in the future. The shock is |$\delta=\exp(1\%)$|⁠. Suppose the average government surplus is 2% of the GDP, then this shock increases the government surplus by about 0.02% of the GDP. By Corollary 1, the Home real exchange rate appreciates by |$1$|%.

Then, by Proposition 1 and assuming |$\alpha\rightarrow 1$|⁠, the Home real exchange rate appreciates by |$\Delta s_t / (A \bar c) \approx 1\%$|⁠, which is the same as the effect of a uniform 1% increase in expected future surpluses. This thought experiment shows that current and expected future surpluses have the same first-order effect on the real exchange rate when they are measured by their present values. Therefore, when I empirically connect exchange rates to government fiscal conditions, it is important to incorporate information about future government surpluses.

2. Explaining Exchange Rate Movements

This section tests the model’s main prediction (Proposition 1 and Corollary 1) that changes in both current and expected future government surpluses should explain concurrent real exchange rate movements. As the model describes countries whose governments do not explicitly default on their debt, I focus on a sample of 11 developed countries: Australia, Canada, Denmark, Germany, Japan, New Zealand, Norway, Sweden, Switzerland, the United Kingdom, and the United States. Internet Appendix B.1 describes the data sources.

2.1 Benchmark results

I develop a measure of current and expected future government surpluses using the government surplus forecasts from the World Economic Outlook Reports. Every April and October from 2009 to 2018, the International Monetary Fund (the IMF) reported its forecasts of each country’s government surpluses in the current year and in the next 5 years. Let |$\mathbb{E}^{IMF}_{t,m}[s^{i}_{t+k}]$| denote the IMF’s forecast of country |$i$|’s government surplus in year |$t+k$|⁠, released in month |$m$| year |$t$|⁠. Since the current year |$t$| has not ended when the forecast is released in month |$m$|⁠, the current year’s government surplus |$\mathbb{E}^{IMF}_{t,m}[s^{i}_{t}]$| is also a forecast.

Note that the ratio between the government surplus and the quantity of outstanding government debt in the previous year measures the net fraction of the outstanding debt the government pays back in the current year. This normalization makes different countries’ government surpluses comparable.

Column 1 in Table 1, panel a, reports the result. An increase in country |$i$|’s government surplus forecast-to-debt ratio relative to country |$j$|’s is associated with real exchange rate appreciation of country |$i$| against country |$j$|⁠. This forecast variable explains 6.0% of the variation in the real exchange rate movement. The coefficient’s magnitude is economically meaningful: a one-standard-deviation increase in the bilateral difference in government surplus forecasts |$(\Delta s^{IMF,i}_{t,m} - \Delta s^{IMF,j}_{t,m})$|⁠, which is about 3.1%, is associated with real exchange rate appreciation of 1.5%. In comparison, the standard deviation of the semiannual real exchange rate movements is 6.8%.

Columns 2 and 3 in Table 1, panel a, report the results. The government surplus forecast-to-debt ratio has a similar explanatory power for the nominal exchange rate movement, but it does not explain the inflation differential. So, the real and the nominal exchange rates have almost the same movements when the government fiscal forecasts change, while the inflation differential does not respond.

This pattern is consistent with the model’s prediction when the home bias in consumption |$\alpha$| approaches 1. In the data, the home bias in consumption is large. In the World Input-Output Database, from 1974 to 2014, the average spending share on domestic goods in consumption (i.e., in final use) was 92%.

2.2 Horse races

Table 1, panel b, reports the results. The IMF’s government surplus forecasts robustly explain concurrent real exchange rate movements in these bivariate regressions, whereas the control variables have weak explanatory power for the real exchange rate movements after the government surplus forecasts are controlled for.

2.3 Robustness

The IMF database provides direct forecasts of government surpluses. It is, however, limited by its short sample period. Moreover, whether the IMF uses exchange rate data to produce their forecasts is unclear. According to IMF, ,it uses a ‘bottom-up’ approach, in which country teams within the IMF generate forecasts for individual countries. These are then aggregated, and through a series of iterations where the aggregates feed back into individual countries’ forecasts, forecasts converge to the projections. Because forecasts are made by the individual country teams, the methodology can vary from country to country and series to series depending on many factors.”¹

I consider two alternative approaches to test the relationship between government surplus forecasts and exchange rate movements in a longer sample. I obtain quarterly government primary surpluses and government debt data since 1980 from Oxford Economics. The sample is unbalanced, but there is no missing observation after each country’s time series starts.

2.3.1 Engel and West (2005) approach

First, I run the ,reverse regression” in Engel and West (2005). If the exchange rate is driven by the expectation of a certain variable, then the exchange rate should be able to predict this variable. In my context, since the real exchange rates are determined by current and expected future government surpluses, real exchange rate movements should be able to predict changes in future government surpluses.

Table 2, panel a, reports the result. The |$\beta^{(k)}$| coefficient is positive and statistically significant when the forecast horizon |$k$| is one or two quarters. Beyond two quarters in the future, the coefficient is not statistically significant, but its magnitude remains similar. So, when currency |$i$| appreciates against currency |$j$|⁠, the future government surpluses in country |$i$| tend to be higher than those in country |$j$|⁠. This reduced-form result is consistent with my model’s prediction that real exchange rates contain information about future fiscal conditions.

2.3.2 VAR approach

Table 2, panel b, reports the results. Consistent with the results in Table 1 that uses the IMF forecasts, an increase in a country’s VAR-based expected government surplus is associated with real and nominal appreciation of its exchange rate, whereas it is not correlated with inflation differential.

3. Explaining the Factor Structure in Currency Returns

3.1 Common surplus factor and fiscal exposure

Since not all countries have government surplus data available at the start of the sample, the common surplus factor in each quarter is the equal-weighted average over countries that have data in that quarter.

Table 3 reports the |$b^i$| coefficients in the two samples. All countries have positive loadings on the common surplus factor, and most loadings are statistically significant. Although the loadings vary across countries, the common surplus factor explains large fractions of variations in many countries. For example, in the Oxford Economics sample, the coefficient for Switzerland is 0.30, while the coefficient for Norway is 1.45, but the common surplus factor explains about the same fraction of variations in both countries.

Internet Appendix B.3 conducts a principal component analysis to confirm the existence of a common factor in the fiscal shocks across countries. For example, in the Oxford Economics sample, the first principal component explains 43% of the variation in the fiscal shock |$\Delta s^i_t$|⁠, and it is highly correlated with the common surplus factor |$f_{t}$| (averaged across all countries) with a correlation of 0.95.

3.2 Explaining the factor structure in currency returns

Having established the existence of the common fiscal factor, I test the prediction that each country’s government fiscal exposure explains its currency’s risk exposure. Previous literature has identified the carry trade return as one of the common risk factors in currency returns (Lustig Roussanov, and Verdelhan 2011, Hassan and Mano 2019, Verdelhan 2018). Accordingly, I measure the risk exposure in currency returns using each currency’s carry trade beta.

Figure 1 plots each country’s currency carry trade beta |$\beta^i_{carry}$| against its fiscal exposure as measured by the regression coefficient |$b^{i}$| in Equation (14). In both samples, fiscal exposures are positively correlated with currency carry betas. For example, Australia has a high fiscal exposure and a high carry trade beta, whereas Japan has a low fiscal exposure and a low carry trade beta.

Table 4, panel a, reports the results. The relationship between the fiscal exposure |$b^i$| and the currency carry trade beta |$\beta_{carry}^{i}$| is economically significant in both samples. A one-standard-deviation increase in a country’s government fiscal exposure is associated with a 0.40 increase in the carry trade beta in the IMF sample, and with a 0.31 increase in the carry trade beta in the Oxford Economics sample. The factor structure in currency returns indeed aligns with the factor structure in government surpluses.

Table 4, panel b, reports the results. After I account for the estimation error in the explanatory variable, the point estimate of |$\lambda$| becomes even larger: a higher government fiscal exposure |$b^{i}$| is associated with a higher carry trade beta |$\beta_{carry}^{i}$|⁠.

These results show that the factor structure in government fiscal conditions aligns with the factor structure in currency returns. This common factor structure also manifests itself elsewhere. For example, Colacito et al. (2018) show that different countries’ GDP growth rates have heterogeneous loadings on the global long-run risk, and these loadings also align with the factor structure in currency returns.

4. Conclusion

In this paper, I develop a model of real exchange rates that attributes a central role to the government budget condition, which requires the discount rate to react endogenously in response to fiscal shocks when prices are sticky. The model predicts that government fiscal conditions explain real exchange rate movements, and the factor structure in fiscal shocks aligns with the factor structure in exchange rate movements. Both predictions are confirmed in the sample of developed countries. These findings help bridge the gap between exchange rates and economic fundamentals.

My model is highly stylized as it features log utility and i.i.d. shocks. In a model with richer preferences, fiscal shocks can also affect the real exchange rate via the risk premium component of the discount rate. Investigating the joint dynamics of fiscal shocks, real exchange rates, and risk premia offers a promising direction for future research.

Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

Footnotes

References