Non-Linearities and the Euler Equation: Does Uncertainty Have an Effect on the Approximation Quality?

Tobias Kranz

doi:10.1515/roe-2019-0031

Published by De Gruyter Oldenbourg January 9, 2020

Non-Linearities and the Euler Equation: Does Uncertainty Have an Effect on the Approximation Quality?

Tobias Kranz

From the journal Review of Economics

https://doi.org/10.1515/roe-2019-0031

Showing a limited preview of this publication:

Abstract

Deriving a forward-looking Euler equation, this paper compares two fully identified non-linear versions. The difference (or bias) between them stems from an approximation by extracting parameters from the expectation values (Jensen’s inequality) as it is common practice in the literature. Furthermore, the model is completely identified using Consensus Forecasts data for the expectations, inflation-indexed bonds as a proxy for the long-run real interest rate, and estimates for the elasticity of intertemporal substitution. Regression analyses using data for three major economies reveal that the difference between the two Euler versions can be explained by uncertainty in the data itself and external uncertainty measures. The results confirm a connection between theoretical and empirical higher-order moments in economic models.

Keywords: consensus forecasts; consumption growth; expectations; interest rates; intertemporal optimization; Jensen’s inequality

JEL Classification: C02; C12; E31; E43

Acknowledgements

Thanks to Christian Bauer, Matthias Neuenkirch, and participants of the 2019 Workshop on Monetary and Financial Macroeconomics at Deutsche Bundesbank Hamburg for helpful comments on earlier versions of the paper.

Appendix

A.1 Euler Equation – Calculation Steps

The optimization problem uses the constraint

(10)Ct⋅Pt+Bt+1=Wt+(1+it−1)⋅Bt,

with W_t as the nominal wage and B_t as the nominal value of bonds. The latter provides the link between two periods. Depending on the definition of the interest rate, the horizon can vary. Since the right-hand side of eq. (10) represents the disposable income in period t, B_t is the investment by households starting in period t - 1 by the interest-bearing condition i_t - 1.

Dynamic Programming uses the additively separable utility function and the envelope theorem to set up optimality conditions for two consecutive periods. The procedure can be divided into three parts. The first part is to write a value function, the Bellman equation. Under the assumption that the second term of the expanded utility

(11)U(Ct)+Et∑s=t+1∞ϱss−t−1U(Cs)

is maximized in period t, the Bellman equation is

(12)V(Bt)≡maxCtU(Ct)+ϱtV(Bt+1).

The expected value vanishes since B_t + 1 is determined by variables in period t in the constraint. Differentiating with respect to C_t gives the first-order condition

(13)dU(Ct)dCt+ϱtdV(Bt+1)dBt+1dBt+1dCt=0,

which results, when including eq. (10), in

(14)dU(Ct)dCt=PtϱtdV(Bt+1)dBt+1.

Equation (14) relates the marginal utility to the marginal value in the following period, the time preference, and prices in the same period. Therefore, a higher ϱ_t and P_t results in a lower C_t.

In the next part, the envelope theorem is used to differentiate the value function (by inserting the optimized Ct∗) with respect to the costate variable B_t:

(15)V(Bt)=U(Ct∗)+ϱtV(Bt+1)

(16)⇒dV(Bt)dBt=ϱtdV(Bt+1)dBt+1⋅dBt+1dBt

(17)⇔dV(Bt)dBt=ϱtdV(Bt+1)dBt+1⋅(1+it−1).

Equation (17) reveals the relationship of the marginal value functions.

In a third step, the first-order condition eq. (14) can be used to replace the value functions in eq. (17) with the marginal utility in both periods t and t - 1:

(18)dU(Ct−1)dCt−11Pt−1ϱt−1=ϱt⋅dU(Ct)dCt1Ptϱt⋅(1+it−1)

(19)⇒dU(Ct)dCt=ϱt⋅(1+it)⋅EtPtPt+1dU(Ct+1)dCt+1.

The time shift yields the Euler condition.

A.2 Constant Relative Risk Aversion – Utility Function

The CRRA-function,

(20)U(C)=C1−γ−11−γ, γ>0,

transforms into a root-, log-, or inverse hyperbolic-form, depending on the value of γ. For γ∈ ]0,1[, a root-function appears, for example the square root:

(21)U(C | γ=0.5)=2(C−1).

Log-utility (γ = 1) can be considered as a special case since l’Hôpital’s rule is needed to reveal the ratio between the limit values in numerator and denominator:

(22)U(C | γ=1)=limγ→1C1−γ−11−γ=limγ→1log(C)⋅C1−γ⋅(−1)−1=log(C).

Finally, a hyperbolic-form, mirrored at the abscissa, emerges for γ > 1:

(23)U(C | γ=2)=C−1−1−1=−1C+1.

Note that for all cases, a positive utility is only assured if more than one unit is consumed.

A.3 Covariances of Transformed Random Variables

The idea is to show the substantial difference in the correlation after transforming two random variables (RV’s). Figure 6 compares the correlation of repeated draws between two (i) normally-distributed and (ii) the inverse of normal-distributed RV’s. Inverting the variables as transformation is chosen because it is relatively simple and basically matches the non-linearities in eq. (5). Per repetition, 30 RV’s are drawn since this roughly corresponds to the number of observations per cross-section, presented in Section 3. We calculate the correlation as standardized covariance, otherwise outliers would distort the density function.

Figure 6:

Resulting correlation with 10⁵ repetitions, each with 30 draws. Distributions: Standard-Normal (left) and inverted Standard-Normal (right).

While the second moments are equal (Std.Dev. = 0.2), the kurtosis shows a fundamental difference: 2.8 (non-transformed) vs. 8.1 (transformed).

A.4 Euro Area – Weighting Scheme

Weighting of EA countries (France, Germany, Italy, Netherlands, and Spain) is conducted from January 1999 to November 2002. Seasonal- and calendar-adjusted real GDP data from Eurostat show that these five countries contribute around 86% to the EA’s GDP. Over time, there is a slight downward trend and a minor structural break in 2001 when Greece joined the euro zone. Quarterly GDP data are converted to a monthly basis via linear interpolation. The weighting scheme is as follows:

(24)μEA=∑i=15κiE[Yi]=E[λY],

where κ_i are GDP-weights with ∑i=15κi=1 and λ, the firm level weights for each country (depending also on number of firms: λi=(N/Ni)⋅κi), is needed to insert the individual forecasts directly into the expectation value, i. e. E[(1+λY)−γ].

Using this method for the period after 11/2002 shows no substantial difference, when comparing with the actual values for the EA.

A.5 Descriptive Statistics

Table 6:

Consumption and price level measures are represented as one-year ahead expected growth rates.

Euler Equation Data	Mean (%)	Std.Dev. (pp)	Min. (%)	Max. (%)
Personal Consumption (US)	2.52	0.87	−1.17	4.49
Household Consumption (UK)	1.92	1.08	−1.95	3.97
Private Consumption (EA)	1.31	0.84	−0.53	2.88
Consumer Prices (US)	2.46	0.87	−0.66	5.2
Retail Prices (UK)	3.01	1.14	0.44	7.55
Consumer Prices (EA)	1.62	0.53	−0.02	3.01
Effective FFR (US)	3.07	2.55	0.04	9.52
SONIA (UK)	3.02	2.53	0.16	8.61
EONIA (EA)	1.7	1.7	−0.36	5.16
Inflation-Indexed Security (US)	1.07	0.89	−0.79	3.14
Inflation-Indexed Bond (UK)	1.82	1.75	−1.68	5.08
Gov. Benchmark Bond Yield (EA)	2.74	1.98	−1	7.02

Table 7:

Consensus Forecasts uncertainty measures are based on the one-year ahead expected growth rates. The last column shows the coefficient of variation.

CF Uncertainty	Mean	Std.Dev.	Min.	Max.	CV
GDP Std.Dev. (US)	0.3	0.11	0.13	0.7	0.37
GDP Std.Dev. (UK)	0.37	0.12	0.17	0.76	0.32
GDP Std.Dev. (EA)	0.27	0.14	0.08	0.7	0.51
Consumption Std.Dev. (US)	0.29	0.1	0.11	0.77	0.34
Consumption Std.Dev. (UK)	0.48	0.17	0.18	1.12	0.36
Consumption Std.Dev. (EA)	0.31	0.15	0.12	0.7	0.49
Prices Std.Dev. (US)	0.28	0.1	0.11	0.92	0.35
Prices Std.Dev. (UK)	0.33	0.17	0.1	1.12	0.5
Prices Std.Dev. (EA)	0.24	0.18	0.06	0.77	0.76

Table 8:

The last column shows the coefficient of variation, only differing by a small amount across related variables. CLIFS is an index of financial stability by the ECB. EPU and MPU are the Baker, Bloom, and Davis (2016) uncertainty measures. Macroeconomic and Financial Uncertainty are the indices by Jurado, Ludvigon, and Ng (2015). Systemic Stress is an indicator provided by the ECB. Originally, the Forecast Uncertainty data by Rossi and Sekhposyan (2015); Rossi and Sekhposyan (2017) are standardized such that the maximum is 1. Slightly larger values in our data are due to interpolation. Interestingly, there is no or weak negative correlation between these measures. The correlation between all other related variables across regions is as expected: positive and significant (available on request). The World Uncertainty Index comes from the paper by Ahir, Bloom, and Furceri (2018).

Uncertainty Measures	Mean	Std.Dev.	Min.	Max.	CV
S&P 500 Std.Dev. (US)	18.85	14.41	1.42	111.41	0.76
FTSE 100 Std.Dev. (UK)	79.76	49.35	11.09	335.78	0.62
Euro Stoxx 50 Std.Dev. (EA)	55.37	40.31	4.75	245.33	0.73
S&P Avg. Daily Range (US)	1.24	0.69	0.37	6.62	0.55
FTSE Avg. Daily Range (UK)	1.28	0.67	0.53	5.95	0.52
EStoxx Avg. Daily Range (EA)	1.69	0.84	0.6	5.88	0.5
VIX (US)	19.29	7.42	9.51	59.89	0.38
VFTSE (UK)	19.19	7.68	9.55	54.15	0.4
VSTOXX (EA)	23.93	8.81	11.99	61.34	0.37
WTI Std.Dev. (US)	1.64	1.35	0.15	11.29	0.82
Brent Std.Dev. (UK/EA)	1.64	1.31	0.16	10.79	0.8
St.Louis FSI (US)	0	1	−1.53	4.71	−
CLIFS (UK)	0.12	0.09	0.01	0.56	0.78
Weighted CLIFS (EA)	0.12	0.08	0.03	0.42	0.66
EPU (US)	113.79	43.05	44.78	284.25	0.38
EPU (UK)	119.66	68.77	24.04	558.22	0.57
EPU (EA)	131.35	61.48	45.3	433.28	0.47
MPU (US)	106.14	50.84	12.8	362.44	0.48
Macro Uncertainty (US)	0.91	0.05	0.85	1.15	0.05
Financial Uncertainty (US)	0.98	0.05	0.91	1.13	0.05
Systemic Stress (EA)	0.18	0.16	0.02	0.8	0.92
Forecast Uncertainty (US)	0.73	0.13	0.49	1.01	0.17
Forecast Inf. Uncertainty (EA)	0.75	0.14	0.45	1.02	0.18
Forecast GDP Uncertainty (EA)	0.75	0.14	0.51	1	0.19
World Uncertainty Index	122.63	37.67	63.76	250.52	0.31

A.6 Correlation – Uncertainty Measures

Table 9:

S&P and WTI are the monthly standard deviations. S&P_R are monthly averages of the daily range in percent (S&P 500). FSI is the St. Louis Financial Stress Index. EPU and MPU are the Baker, Bloom, and Davis (2016) uncertainty measures. Macro (Macroeconomic Uncertainty) and Fin. (Financial Uncertainty) are the uncertainty measures by Jurado, Ludvigon, and Ng (2015). Rossi is the forecast uncertainty by Rossi and Sekhposyan (2015). WUI is the World Uncertainty Index by Ahir, Bloom, and Furceri (2018). Numbers in bold denote significance at the 5% level.

US	S&P	S&PR	VIX	WTI	FSI	EPU	MPU	Macro	Fin.	Rossi
S&P	1	−	−	−	−	−	−	−	−	−
S&P_R	0.6	1	−	−	−	−	−	−	−	−
VIX	0.49	0.9	1	−	−	−	−	−	−	−
WTI	0.45	0.45	0.37	1	−	−	−	−	−	−
FSI	0.11	0.68	0.67	0.08	1	−	−	−	−	−
EPU	0.45	0.39	0.39	0.32	−0.04	1	−	−	−	−
MPU	0.23	0.07	0.06	−0.04	−0.08	0.48	1	−	−	−
Macro	0.36	0.63	0.6	0.59	0.55	0.23	−0.06	1	−	−
Fin.	0.46	0.76	0.83	0.31	0.64	0.3	−0.02	0.65	1	−
Rossi	0.03	0.14	0.19	−0.08	0.17	0.16	0.05	0.1	0.18	1
WUI	0.13	–0.28	–0.29	0.14	–0.62	0.44	0.37	–0.21	–0.36	0.08

Table 10:

FTSE and Brent are the monthly standard deviations. FSTE_R are monthly averages of the daily range in percent (FTSE 100). CLIFS is an index of financial stability by the ECB. EPU is the Baker, Bloom, and Davis (2016) uncertainty measure. WUI is the World Uncertainty Index by Ahir, Bloom, and Furceri (2018). Numbers in bold denote significance at the 5% level.

UK	FSTE	FSTER	VFTSE	Brent	CLIFS	EPU
FSTE	1	−	−	−	−	−
FSTE_R	0.76	1	−	−	−	−
VFTSE	0.65	0.93	1	−	−	−
Brent	0.39	0.43	0.28	1	−	−
CLIFS	0.38	0.56	0.6	0.32	1	−
EPU	0.2	0.21	0.13	0.17	0.18	1
WUI	0	−0.11	–0.26	0.18	–0.18	0.59

Table 11:

EStoxx and Brent are the monthly standard deviations. EStoxx_R are monthly averages of the daily range in percent (ESTOXX 50). CLIFS is an index of financial stability by the ECB. EPU is the Baker, Bloom, and Davis (2016) uncertainty measure. INF and GDP are forecast error measures by Rossi and Sekhposyan (2017). WUI is the World Uncertainty Index by Ahir, Bloom, and Furceri (2018). Numbers in bold denote significance at the 5% level.

EA	EStoxx	EStoxxR	VStoxx	Brent	CLIFS	EPU	CISS	INF	GDP
EStoxx	1	−	−	−	−	−	−	−	−
EStoxx_R	0.7	1	−	−	−	−	−	−	−
VStoxx	0.62	0.92	1	−	−	−	−	−	−
Brent	0.32	0.2	0.18	1	−	−	−	−	−
CLIFS	0.4	0.72	0.75	0.31	1	−	−	−	−
EPU	0.15	0.15	0.13	0.29	0.14	1	−	−	−
CISS	0.37	0.59	0.61	0.48	0.79	0.15	1	−	−
INF	−0.12	–0.21	–0.21	0.31	−0.05	0.11	0.06	1	−
GDP	0.27	0.24	0.25	0.27	0.28	−0.12	0.36	−0.1	1
WUI	−0.1	–0.13	–0.14	0.18	–0.16	0.78	−0.12	0.18	–0.38

Table 12:

Symbols.

Letter	Description
β_i	Parameters (“bias” equation, i∈{0,1,2})
γ_τ	Reciprocal value of the EIS (τ∈{1,2})
Δ_i	Residuals (Jensen’s inequality, i∈{c,π})
εt(i)	Error terms (i∈{J,bias})
κ_i	(Aggregated) GDP weights (i∈{1,…,5})
λ_i	GDP weights on individual level (i∈{1,…,5})
µ	Mean
π_t	Inflation
ϱ_t	Discount factor (time preference)
σ	Standard deviation
B_t	Bonds
c_t	Consumption growth
C_t	Consumption
i_t	Nominal interest rate
m	Index of months
N	Number of forecasters
N_i	Number of forecasters in a specific country (i∈{1,…,5})
P_t	Price; Price level
r_t	Long-run real interest rate
U(.)	Utility function
V(.)	Value function (Bellman equation)
W_t	Wage

References

Ahir, H., N. Bloom and D. Furceri (2018): The World Uncertainty Index, SSRN 3275033. https://doi.org/10.2139/ssrn.3275033. Search in Google Scholar

Attanasio, O. P. and H. Low (2004): Estimating Euler Equations, Review of Economic Dynamics 7, 406–435. https://doi.org/10.1016/j.red.2003.09.003.Search in Google Scholar

Attanasio, O. P. and G. Weber (1989): Intertemporal Substitution, Risk Aversion and the Euler Equation for Consumption, The Economic Journal 99, 59–73. https://doi.org/10.2307/2234070.Search in Google Scholar

Baker, S., N. Bloom and S. J. Davis (2016): Measuring Economic Policy Uncertainty, The Quarterly Journal of Economics 131, 1593–636. https://doi.org/10.1093/qje/qjw024.Search in Google Scholar

Beaudry, P. and E. van Wincoop (1996): The Intertemporal Elasticity of Substitution: An Exploration using a US Panel of State Data, Economica 63, 495–512. https://doi.org/10.2307/2555019.Search in Google Scholar

Blanchard, O. J. and N. G. Mankiw (1988): Consumption: Beyond Certainty Equivalence, American Economic Review 78, 173–177. https://jstor.org/stable/1818118.10.3386/w2496Search in Google Scholar

Carroll, C. D. (2001): Death to the Log-Linearized Consumption Euler Equation! (And Very Poor Health to the Second-Order Approximation), Advances in Macroeconomics 1, Article 6. https://doi.org/10.2202/1534-6013.1003.Search in Google Scholar

Carroll, C. D. and M. S. Kimball (1996): On the Concavity of the Consumption Function, Econometrica 64, 981–992. https://doi.org/10.2307/2171853. .Search in Google Scholar

Epstein, L. G. and S. E. Zin (1989): Substitution, Risk Aversion, and the Temporal Behavior of Consumption and Asset, Econometrica 57, 937–969. https://doi.org/10.2307/1913778.Search in Google Scholar

Galí, J. (2015): Monetary Policy, Inflation, and the Business Cycle. Princeton University Press, Princeton, NJ, 2nd ed. https://press.princeton.edu/titles/10495.html.Search in Google Scholar

Gomes, F. A. R. and L. S. Paz (2013): Estimating the Elasticity of Intertemporal Substitution: Is the Aggregate Financial Return free from the Weak Instrument Problem? Journal of Macroeconomics 36, 63–75. https://doi.org/10.1016/j.jmacro.2013.01.005 .Search in Google Scholar

Gürkaynak, R. S., B. Sack and J. H. Wright (2010): The TIPS Yield Curve and Inflation Compensation, American Economic Journal: Macroeconomics 2, 70–92. https://doi.org/10.1257/mac.2.1.70.Search in Google Scholar

Hall, R. E. (1978): Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence, Journal of Political Economy 86, 971–987. https://jstor.org/stable/1840393.10.1086/260724Search in Google Scholar

Hall, R. E. (1988): Intertemporal Substitution in Consumption, Journal of Political Economy 96, 339–357. https://jstor.org/stable/1833112.Search in Google Scholar

Hamilton, J. D., E. S. Harris, J. Hatzius and K. D. West (2016): The Equilibrium Real Funds Rate: Past, Present, and Future, IMF Economic Review 64, 660–707. https://doi.org/10.1057/s41308-016-0015-z.Search in Google Scholar

Havránek, T., R. Horvath, Z. Irsova and M. Rusnak (2015): Cross-Country Heterogeneity in Intertemporal Substitution, Journal of International Economics 96, 100–118. https://doi.org/10.1016/j.jinteco.2015.01.012.Search in Google Scholar

Jensen, J. L. W. V. (1906): Sur les Fonctions Convexes et les Inégalités Entre les Valeurs Moyennes, Acta Mathematica 30, 175–193. https://doi.org/10.1007/BF02418571.Search in Google Scholar

Jurado, K., S. C. Ludvigon and S. Ng (2015): Measuring Uncertainty, American Economic Review 105, 1177–1216. https://doi.org/10.1257/aer.20131193. Search in Google Scholar

Kydland, F. E. and E. C. Prescott (1982): Time to Build and Aggregate Fluctuations, Econometrica 50, 1345–1370. https://doi.org/10.2307/1913386 .Search in Google Scholar

Laubach, T. and J. C. Williams (2003): Measuring the Natural Rate of Interest, The Review of Economics and Statistics 85, 1063–1070. https://jstor.org/stable/3211826.Search in Google Scholar

Lettau, M. and S. C. Ludvigson (2009): Euler Equation Errors, Review of Economics Dynamics 12, 255–283. https://doi.org/10.1016/j.red.2008.11.004.Search in Google Scholar

Lovell, M. C. (1986): Tests of the Rational Expectations Hypothesis, American Economic Review 76, 110–124. https://jstor.org/stable/1804130.Search in Google Scholar

Lucas, R. E. (1976): Econometric Policy Evaluation: A Critique, Carnegie-Rochester Conference Series on Public Policy 1, 19–46. https://doi.org/10.1016/S0167-2231(76)80003-6.Search in Google Scholar

Ludvigson, S. C. and C. H. Paxson (2001): Approximation Bias in Linearized Euler Equations, Review of Economics and Statistics 83, 242–256. https://doi.org/10.1162/00346530151143789.Search in Google Scholar

Newey, W. K. and K. D. West (1987): A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix, Econometrica 55, 703–708. https://doi.org/10.2307/1913610.Search in Google Scholar

Pratt, J. W. (1964): Risk Aversion in the Small and in the Large, Econometrica 32, 122–136. https://doi.org/10.2307/1913738.Search in Google Scholar

Rossi, B. and T. Sekhposyan (2015): Macroeconomic Uncertainty Indices Based on Nowcast and Forecast Error Distributions, American Economic Review 105, 650–655. https://doi.org/10.1257/aer.p20151124.Search in Google Scholar

Rossi, B. and T. Sekhposyan (2017): Macroeconomic Uncertainty Indices for the Euro Area and its Individual Member Countries, Empirical Economics 53, 41–62. https://doi.org/10.1007/s00181-017-1248-z.Search in Google Scholar

Schmitt-Grohé, S. and M. Uribe (2004): Solving Dynamic General Equilibrium Models Using a Second-Order Approximation to the Policy Function, Journal of Economic Dynamics & Control 28, 755–775. https://doi.org/10.1016/S0165-1889(03)00043-5.Search in Google Scholar

Zeldes, S. P. (1989): Optimal Consumption with Stochastic Income: Deviations from Certainty Equivalence, The Quarterly Journal of Economics 104, 275–298. https://doi.org/10.2307/2937848.Search in Google Scholar

Published Online: 2020-01-09

Published in Print: 2020-01-28

From the journal

Review of Economics

Volume 70 Issue 3

Journal and Issue

Articles in the same Issue

Frontmatter

Research Articles

Some Reflections on the Political Economy of Monetary Policy

The Impact of Quantitative Easing on Bank Loan Supply and Monetary Policy Implementation in the Euro Area

Non-Linearities and the Euler Equation: Does Uncertainty Have an Effect on the Approximation Quality?