Fiscal policy in an emerging market business cycle model

TL;DR: In this paper, the authors reconcile the procyclicality of real interest rates with the above facts by embedding fiscal policy into a standard emerging market business cycle model, and show that fiscal policy makes real interest rate a-cyclical or pro cyclical, and use the model to replicate some of the key features of the Indian business cycle.

Abstract: Emerging market economy business cycles are typically characterized by high consumption and output volatility, strongly counter-cyclical current accounts, and counter-cyclical real interest rates. Evidence from the wider EME and less developed economy business cycle experience suggests however that real interest rates can also be pro-cyclical. We reconcile the pro-cyclicality of real interest rates with the above facts by embedding fiscal policy into a standard emerging market business cycle model. We show that fiscal policy makes real interest rates a-cyclical or pro-cyclical. We use the model to replicate some of the key features of the Indian business cycle.

Summary

1 Introduction

• Building dynamic stochastic general equilibrium (DSGE) models of emerging market economies (EMEs) has become an important area of research in macroeconomics.
• These authors explain the key stylized facts of EME business cycles discussed above by allowing for both permanent trend shocks and transitory changes in productivity.
• The authors takeaway from this literature is that in some EMEs, government expenditures are counter-cyclical with respect to output and in others it is procyclical.
• Over the last decade, however, several EMEs have "graduated" from having pro-cyclical scal policy to having counter-cyclical scal policy.

1.1 Description and Main Results

• The authors develop a small open economy (SOE) real business cycle (RBC) model along the lines of Neumeyer and Perri (2005) with two crucial di¤erences.
• The authors assume that the government imposes time invariant distortionary taxes on consumption, labor income and capital income, and maintains a balanced budget at every time period.
• The authors theoretical contribution is two-fold: rst, the authors characterize the scal policy wedge in closed-form under a variety of assumptions on scal 9NP stands for Neumeyer and Perri (2005) and AG stands for Aguiar and Gopinath (2007).
• 10See Male (2010) policy, and show how this a¤ects movements in labor supply adversely; and second, the authors show that because the scal policy wedge is time varying and increases with a positive interest rate shock, the impact of an increase of the wedge on labor supply is higher when there is a higher weight on government consumption in utility.
• Thus, both labor supply and labor demand channels make the real interest rate a-cyclical, and under certain cases, pro-cyclical, matching the qualitative features of the EME data in Table 2.12.

1.1.1 Indian Business Cycle

• The authors choose India because India typi es the broader EME business cycle experience listed in Column 3 (Table 2).
• The key Indian stylized facts are: higher relative consumption volatility, higher relative investment volatility, counter-cyclical net exports, counter-cyclical government expenditures, and a pro-cyclical interest rate (see Ghate et al. (2013), Table 5).
• The counter-cyclicality of government expenditures has been coupled with pro-cyclical interest rates and counter-cyclical next exports, consistent with the evidence on other EMEs reported in Table 1 and Table 2.
• 13We calibrate their model using Dynare Version 4.3.0.the authors.the authors.
• Sector banks still own 70% of the banking sector s assets in India.

2.1 The Firm s Problem

• The economy consists of rms, a government, and households.
• To meet this working capital constraint, the rm borrows from the government and from households by issuing debt.
• The authors assume that the production technology, yt; exhibits constant returns to scale (CRS).
• The timing of events and decisions is given in Figure (1).

2.1.1 Firms Pro t Maximizing Conditions

• The rm s pro t maximization yields the following rst order conditions 8t; for labor, lt; and capital, ekt 1; respectively.
• For given values of and G; interest rate shocks a¤ect wage payments with a lag since effective wage payments depend on RPt 1:.

2.2 Government

• All variables are therefore transformed to their corresponding stationary values except lt; which is assumed to be stationary.
• The interest income is given by, RGt 1 Gwtlt.
• The authors assume that net of Gt; the government lends.

2.3 The Household s Problem

• Each representative household consumes and invests a homogenous good and supplies labor and capital to rms.
• The representative household has the following expected discounted lifetime utility E0 1X t=0 tU(c t ; lt); (8) where 2 (0; 1) denotes the households subjective discount factor.
• Assuming GHH preferences ensures that labor supply is independent of consumption and therefore interest rates.
• The parameter is the coe¢ cient of risk aversion and is the intra-temporal elasticity of substitution of labor supply.
• Higher provision of services elicits a strong reduction in the private consumption of these services.

2.5.1 The e¤ect of interest rate shocks on labor supply

• Note that from equation (16), and unlike the case with GHH preferences, labor supply depends on current levels of e¤ective consumption, ec t because of the income e¤ect, and ewt.
• The above proposition derives su¢ cient conditions under which t >.
• From equation (24), interest rate shocks a¤ect labor supply through two channels in time period, t: A positive interest rate shock causes consumption, ect; to instantaneously fall due to the standard inter-temporal substitution e¤ect (equation (17)).
• Also, for a higher value of ; the increase in the scal policy wedge is higher for a given interest rate shock.
• This causes a larger net increase in labor supply in time period, t: When is small (< 1), the inter-temporal substitution e¤ect has a smaller e¤ect on labor supply than the scal policy wedge.

2.5.2 The e¤ect of interest rate shocks on labor demand

• The presence of the subsidy parameters G and s however dampens the inward shift of lDt+1.
• Therefore from Proposition 2 and Proposition 3 the authors obtain the impact of a single period positive interest rate shock on equilibrium labor and output in time period t.
• This causes output yt to increase on impact, that is, @eyt @RPt > 0: Fiscal policy dampens the movements in equilibrium labor.
• If the scal policy wedge is strong and the subsidy parameters, (s; G) are small, then a downward movement in labor demand will unambiguously decrease full employment output.

3 Calibration

• Based on the quarterly data available on the Indian macroeconomy documented in Ghate et al. (2013), the stylized facts relevant for the Indian economy are25, (a) higher relative consumption volatility, (b) counter-cyclical net exports, (c) counter-cyclical government expenditures, and (d) a pro-cyclical real interest rate.
• These facts are based on quarterly data available on the Indian macroeconomy documented in Ghate et al. (2013), for which the authors seek to replicate qualitatively.
• As the authors noted in the introduction, while the rst two facts are common to a wide variety of EMEs, there is no robust stylized observation on the correlation between real government expenditure and output.
• Also, while Neumeyer and Perri (2005) and Uribe and Yue (2005) state that interest rates are generally counter-cyclical in EMEs, Male (2010) nds this observation not to be universally true particularly among EMEs in Africa, Asia and Eastern Europe.
• The authors theoretical model can therefore be seen as providing a more general model that produces a range of business cycle outcomes that are consistent with the broader EME experience.

3.1 Parameter Values

• The authors x the quarterly capital depreciation rate at = 0:025 which approximately matches the annual depreciation rate in India of 10%: The authors choose = 0:4 from Ghate et al. (2012).
• In their baseline calibrations, the authors arbitrarily set G = 127; which means that the entire working capital constraint is subsidized.
• See Table 170, Handbook of Statistics of the Indian Economy, RBI.
• As the authors will show, this assumption is crucial for making consumption more volatile than output in their model.

3.2 Estimation of the data generating processes

• The authors calibrate the model using total factor productivity (TFP) shocks and interest rate shocks.
• These lending rates are di¤erentiated according to the credit-worthiness of borrowers.

3.3 Single Period Shocks

• There are three shocks in their model TFP shocks.
• BAt , world interest rate shocks bR t , and shocks to the country spread risk (ut).
• The authors will analyze the e¤ect of single period shocks on eyt; ect; ext; eGt; elt; and net exports,fnxt:35 The authors will see how a single period 10% shock a¤ects the deviations of these variables from their corresponding steady state values.

3.3.1 TFP shock

• Figure (6) plots the impulse response functions due to a single period shock in total factor productivity (A).
• As a result, the deviation of output from its corresponding steady state value (ly) increases.
• They also estimate another case, the independent country risk case, where, bDt; is assumed evolve according to an exogenous process.
• Government consumption (lg) also increases on impact rst due to a positive TFP shock.

3.3.2 Interest rate shock

• Figure (7) shows the impact of a single period shock to the world interest rate R :.
• The domestic interest rate (R) increases which causes an instantaneous drop in private consumption (lc) due to the inter-temporal substitution e¤ect.
• This is because an initial reduction in tax revenue from lower consumption levels dominates the increase in tax revenue from the wage and rental income taxation, also known as 1.
• The government revenue to expenditure gap denoted by tr_gp (see (33)) falls because labor (ll) increases.
• The impulse responses however seem to be converging to the steady state very quickly since u is random and not persistent.

3.4 Multi-period Shocks

• Next, the authors calibrate their model with multi-period uncorrelated shocks to TFP bAt , world interest rates bR t and idiosyncratic shocks to the country spread (but) and compare the second order moments of their simulated data with the Indian quarterly data from 1999 Q2 to 2010 Q2.37 Table (4) summarizes their calibration results.
• The results from estimating this model are reported in the column "No Fiscal Policy".
• The scal policy wedge, t = = 1: Second, the authors include only government consumption eGt nanced by factor income taxes.
• The authors report results obtained by estimating this model in the column "G and S".
• The column "G and S (with high )" reports results for a high value of = 75: Finally, the column "Actual Data" reports the actual second order moments of the Indian data from Ghate et al. (2013).

3.4.1 No scal policy

• The labor supply equation (24), becomes the standard labor supply equation in the absence of scal policy: lSt = 1 ectewt 1 : (35) As shown in column 2 in Table 4; the authors observe that consumption, investment, and the netexport to output ratio are pro-cyclical, whereas real interest rates are weakly counter-cyclical.
• While this model under-estimates the relative volatility of consumption and the real interest rate, it over-estimates the relative volatility of investment and the net-export to output ratio.

3.4.2 Government consumption

• Suppose that the government imposes factor income taxes and spends it only on government consumption that a¤ects utility.
• The government, however, does not subsidize working capital loans.
• Therefore any change in consumption and income directly a¤ects government consumption.
• A positive interest rate shock does not have a direct e¤ect on the scal policy wedge, although it does a¤ect the wedge, and therefore lSt ; indirectly through other endogenous variable such as ect; rt; ekt; and ewt: Including eGt also makes the net-exports to output ratio more pro-cyclical.

3.4.3 Government consumption and subsidy

• The moments are summarized in column 4 of Table 4: In contrast to the model with only government consumption, the authors now get counter-cyclical government consumption, counter-cyclical net-exports to output ratio, pro-cyclical consumption and investments, and weakly counter-cyclical real interest rates.
• The authors model qualitatively replicates the standard stylized facts that motivate the theoretical framework of Neumeyer and Perri (2005) through an alternate but compatible mechanism.
• While private investments are less pro-cyclical as compared to the data (which is due to a highly over-estimated investment volatility), interest rates continue to be weakly countercyclical.
• Relative volatility of consumption, investments, net exports, and government expenditures are all closer to the Indian data and higher in the presence of subsidies to the working capital loans to rms.
• Government expenditures are now signi cantly counter-cyclical due to high subsidies given to rms.

3.4.4 Government consumption and subsidy (under high )

• In their model, scal policy a¤ects the economy by distorting the labor market equilibrium.
• The authors therefore analyze the e¤ect of changes in on some of their calibration results.
• The authors model suggests that a higher generates this higher consumption volatility.
• The net e¤ect is that equilibrium labor increases by more due to a positive interest rate shock.
• This makes government expenditures less counter-cyclical with a progression towards becoming pro-cyclical.

4 The Role of Fiscal Policy as an Automatic Stabilizer

• The authors model outlines how scal policy can play the role of an automatic stabilizer when an economy is hit with interest rate shock which adversely a¤ects labor market outcomes and also real output.
• Since government expenditure in their model is non-discretionary, it adjusts automatically with other endogenous variables.
• Table 4 shows how scal policy dampens overall volatility in the economy, but leads to a trade-o¤.
• This makes private consumption s response to a positive interest rate shock high.
• The net e¤ect on labor supply is therefore positive.

5 Conclusion

• The authors build a tractable small open economy RBC model in which scal policy has a role in making pro-cyclical real interest rates consistent with counter-cyclical net exports and higher consumption volatility.
• The authors then calibrate the model to India to qualitatively match its business cycle properties.
• The authors also discuss the role that scal policy as an automatic stabilizer in the context of their model.
• From a policy standpoint, their model suggests how the adverse e¤ects of interest rate shocks on labor market outcomes can be mitigated by scal policy.

Figures

Figure 2: The e¤ect of an increase in RPt on l S t

Table 4 : Comparison between the model and the data38

Table 3: Summary of Parameter Values

Figure 8: Impact of a single period spread bD shock

Figure 6: Impact of a single period TFP bA shock

Figure 7: Impact of a single period international interest rate bR shock

Figure 4: The e¤ect of an increase in RPt on l D t+1

Figure 5: The e¤ect of a subsidy G and s on lDt+1

Figure 3: The e¤ect of the wedge t on lSt

Figure 1: Timing of Events and Decisions

Table 1 : Real interest rates and government expenditures in EMEs fromMale (2010)

Table 2: Facts based on the wider literature on EME business cycles

Frequently asked questions

Q1. What have the authors contributed in "Fiscal policy in an emerging market business cycle model" ?

The authors show that scal policy makes real interest rates a-cyclical or pro-cyclical. 

Q2. What have the authors stated for future works in "Fiscal policy in an emerging market business cycle model" ?

For future work, the authors hope to introduce sovereign debt and endogenize country spreads with sovereign default risks. 

Q3. What is the role of scal policy in the small open economy?

The authors build a tractable small open economy RBC model in which scal policy has a role in making pro-cyclical real interest rates consistent with counter-cyclical net exports and higher consumption volatility. 

Q4. What does the effect of a positive shock to RPt on labor demand?

A positive shock to interest rate RPt lowers labor demand only in time period t+ 1: However, the presence of G and s; dampens the reduction in lDt+1. 

Q5. What is the effect of a positive interest rate shock on the net exports to output ratio?

As increases, with a positive interest rate shock, on the one hand, an increase in income and an increase in the interest rate increases revenue generated from wage, capital, and interest incomes. 

Q6. Why are governments forced to reduce spending during recessions?

During recessions, governments in EMEs are forced to reduce spending because of lack of access to credit (see Talvi and Vegh (2005)). 

Q7. What causes the savings-investment gap to increase on impact?

An increase in savings due to postponement of consumption and reduction in investments causes the savings-investment gap (shown as si_gp) to increase on impact. 

Q8. What is the net exports to output ratio?

The net-exports to output ratio, given bynxt = (st xt) + (TRt Gt)yt ;are counter-cyclical because of a falling savings-investment gap and a negative public revenueexpenditures gap. 

Q9. How do rms nance their working capital constraint?

In order to nance this working capital constraint, rms issue corporate bonds to agents in international capital markets at a market determined interest rate on bonds. 

Q10. what causes a larger net increase in labor supply in time period?

This causes a larger net increase in labor supply in time period, t: When is small (< 1), the inter-temporal substitution e¤ect has a smaller e¤ect on labor supply than the scal policy wedge. 

Q11. What is the correlation coefficient of the variable Z with output Y?

While this model under-estimates the relative volatility of consumption and the real interest rate, it over-estimates the relative volatility of investment and the net-export to output ratio.38 (Z; Y ) is the correlation coe¢ cient of variable Z with output Y: (Z)= (Y ) is the relative standard deviation of variable Z with output Y: Also, refer to table (2) for second order moments of the Indian data.