Download the Full Version of textbook for Fast Typing at textbookfull.

com

Persistent Stochastic Shocks in a New Keynesian

Model with Uncertainty 1st Edition Tobias Kranz
(Auth.)

https://textbookfull.com/product/persistent-stochastic-
shocks-in-a-new-keynesian-model-with-uncertainty-1st-
edition-tobias-kranz-auth/

OR CLICK BUTTON

DOWNLOAD NOW

Download More textbook Instantly Today - Get Yours Now at textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Model Predictive Control of High Power Converters and

Industrial Drives 1st Edition Tobias Geyer

https://textbookfull.com/product/model-predictive-control-of-high-
power-converters-and-industrial-drives-1st-edition-tobias-geyer/

textboxfull.com

Reasoning and Public Health New Ways of Coping with

Uncertainty 1st Edition Louise Cummings (Auth.)

https://textbookfull.com/product/reasoning-and-public-health-new-ways-
of-coping-with-uncertainty-1st-edition-louise-cummings-auth/

textboxfull.com

Learning with Uncertainty 1st Edition Xizhao Wang

https://textbookfull.com/product/learning-with-uncertainty-1st-
edition-xizhao-wang/

textboxfull.com

The persistent objector rule in international law 1st

Edition James A. Green

https://textbookfull.com/product/the-persistent-objector-rule-in-
international-law-1st-edition-james-a-green/

textboxfull.com
Building the Internet of Things A Project Workbook Kranz
Maciej

https://textbookfull.com/product/building-the-internet-of-things-a-
project-workbook-kranz-maciej/

textboxfull.com

Stochastic Finance An Introduction with Examples 1st

Edition Turner

https://textbookfull.com/product/stochastic-finance-an-introduction-
with-examples-1st-edition-turner/

textboxfull.com

Artificial Intelligence with Uncertainty Second Edition Du

https://textbookfull.com/product/artificial-intelligence-with-
uncertainty-second-edition-du/

textboxfull.com

The Sociology of Arts and Markets: New Developments and

Persistent Patterns Andrea Glauser

https://textbookfull.com/product/the-sociology-of-arts-and-markets-
new-developments-and-persistent-patterns-andrea-glauser/

textboxfull.com

Business Cycle Dynamics and Stabilization Policies A

Keynesian Approach 1st Edition Hajime Hori (Auth.)

https://textbookfull.com/product/business-cycle-dynamics-and-
stabilization-policies-a-keynesian-approach-1st-edition-hajime-hori-
auth/
textboxfull.com
Tobias Kranz

Persistent Stochastic
Shocks in a New
Keynesian Model
with Uncertainty
BestMasters
Springer awards „BestMasters“ to the best master’s theses which have been com-
pleted at renowned Universities in Germany, Austria, and Switzerland.
The studies received highest marks and were recommended for publication by
supervisors. They address current issues from various fields of research in natural
sciences, psychology, technology, and economics.
The series addresses practitioners as well as scientists and, in particular, offers guid-
­ance for early stage researchers.
Tobias Kranz

Persistent Stochastic
Shocks in a New
Keynesian Model
with Uncertainty
Tobias Kranz
Trier, Germany

Master Thesis, University of Trier 2015

BestMasters
ISBN 978-3-658-15638-1 ISBN 978-3-658-15639-8 (eBook)
DOI 10.1007/978-3-658-15639-8

Library of Congress Control Number: 2016954031

Springer Gabler
© Springer Fachmedien Wiesbaden 2017
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained
herein or for any errors or omissions that may have been made.

Printed on acid-free paper

This Springer Gabler imprint is published by Springer Nature

The registered company is Springer Fachmedien Wiesbaden GmbH
The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany
Acknowledgements
I would like to take this opportunity to thank several people who have supported me
during the two years of my master studies and been of great help in the work on this
thesis.
I highly value all the input Jun. Prof. Dr. Matthias Neuenkirch has given me. He
answered every one of my questions swiftly and thoughtfully. I admire his work ethics and
thank him for guiding me through this process in such a supportive manner. Furthermore,
I would like to thank Stefan Geisen for his great mathematical help in past projects and
for the many fruitful conversations we had on the topics of economics and mathematics.
I would also like to express my gratitude to Sarah Cames for developing and improving
my English skills and for being so diligent in correcting my mistakes.
Finally, I wish to thank my family for their endless support and encouragement, not
only throughout my studies, but throughout my whole life.

Tobias Kranz
Trier, November 2015
Contents
Lists of Figures and Tables IX

Lists of Symbols and Acronyms XI

1 Introduction 1

2 New Keynesian Model 3

2.1 The New Neoclassical Synthesis . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 New Keynesian Phillips Curve . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Forward-Looking IS Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Targeting Rule under Discretion . . . . . . . . . . . . . . . . . . . . . . . . 17

3 Persistent Shocks 20
3.1 AR(1) Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Uniqueness of an Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4 Comparative Static Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.5 Parameter Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.6 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4 Model with Uncertainty 32

4.1 Quadratic Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.2 Interest Rate Rule with the Quadratic IS Curve . . . . . . . . . . . . . . . 35
4.3 Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.4 Further Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.5 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.6 Model Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

5 Conclusion 49

References 51

Appendix 57
List of Figures
1 Graphical Results of Households’ and Firms’ Static Optimization . . . . . 9
2 Section 3.6: Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . 29
3 Section 3.6: Strict Inflation Targeting . . . . . . . . . . . . . . . . . . . . . 31
4 The Quadratic IS Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5 Section 4.5: Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . 41
6 Section 4.5: Strict Inflation Targeting . . . . . . . . . . . . . . . . . . . . . 42
7 Section 4.5: No Persistence . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8 Section 4.6: Equilibrium Condition, Differences . . . . . . . . . . . . . . . 44
9 Section 4.6: Equilibrium Condition, Differences with µ = 0.8 . . . . . . . . 44
10 Section 4.6: Differences of it in Several Scenarios (µ variable) . . . . . . . . 46
11 Section 4.6: Differences of it in Several Scenarios (κ variable) . . . . . . . . 46
A1 Linear NKPC and NKPC without Log-Linearization . . . . . . . . . . . . 62
A2 Differences after Log-Linearization of the NKPC . . . . . . . . . . . . . . . 63
A3 IS Curve without Log-Linearization and an Investment Trap Type Case . . 66
A4 Appendix A.17: Strict Inflation Targeting, Differences . . . . . . . . . . . . 70
A5 Appendix A.17: No Persistence, Differences . . . . . . . . . . . . . . . . . 70
A6 Appendix A.18: Difference of it in Several Scenarios (σ variable) . . . . . . 71
A7 Appendix A.18: Difference of it in Several Scenarios (δ variable) . . . . . . 71
A8 Appendix A.18: Difference of it in Several Scenarios (σe2 variable) . . . . . 72

List of Tables
1 Greek Alphabet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI
2 Latin Alphabet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII
3 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .XIII
4 Overview of all Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5 Overview of all Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Lists of Symbols and Acronyms

Table 1: Greek Alphabet

Letter Description

α Summarizing parameters (αγ , αy , απ , αµ , αe , ασ )

β Discount factor (time preference)
γ Parameter in cost function
δ Weighting on output gap in loss function
ε Elasticity of substitution
ζ Error term of cost shock
η Error term of demand shock
θ Auxiliary parameter
κ Slope of NKPC
λ Lagrange variable (multiplier); Eigenvalue
µ Cost shock persistence
ν Demand shock persistence
ξ Integration variable
π Inflation

σ Reciprocal value of the IES

σe2 Cost shock variance
σu2 Demand shock variance
τ Firm index

φ Price stickiness
χ Lagrange variable (multiplier)
ψ Lagrange variable (multiplier)
 Table 2: Latin Alphabet

Letter Description

c Constants in cost function (cf ix , cvar )

e Cost shock
i Nominal interest rate; Imaginary number
j Control variable
k Cost parameter in Calvo pricing
n Natural number
p Log-linearized price around the steady state
q Parameter in the generalized mean formula
r Long-run real interest rate
s Control variable in intertemporal optimizations
ss Notation for steady state (long-term value)
t Time variable
u Demand shock
x Calvo price
y Log-linearized output growth rate around the steady state
y Growth rate of output gap around the steady state
y Output growth rate
z Example for a log-linearized variable
A Vectors in a system of difference equations (A0 , At , At+1 )
B Bonds
C Consumption
K(.) Cost function
L (.) Lagrange function
M Coefficient matrix
P Price; Price level
U (.) Utility function
V (.) Value function in Bellman equation
W Wage
Y Output
Z Example variable

XII
 Table 3: Abbreviations

Abbr. Term

AD Aggregated Demand
AR Autoregressive
AS Aggregated Supply
BL Baseline
CES Constant Elasticity of Substitution
DSGE Dynamic Stochastic General Equilibrium
ECB European Central Bank
GDP Gross Domestic Product
IES Intertemporal Elasticity of Substitution
IS Investment/Saving
MATLAB Matrix Laboratory
NKM New Keynesian Model
NKPC New Keynesian Phillips Curve
RBC Real Business Cycle

XIII
1 Introduction
The main objective of this paper is to examine a dynamic general equilibrium condition out
of a basic three-equation New Keynesian model (NKM), augmented with stochastic terms
and non-linearity. More precisely, the additive terms behave like persistent stochastic
shocks and are modeled as an exogenous first-order autoregressive process. In line with
the literature (see, among others, the textbooks by Galı́ 2015 and Walsh 2010) cost shock
and demand shock are utilized for the New Keynesian Phillips curve (NKPC) and the
forward-looking IS curve respectively. Non-linearity enters the model through a second-
order Taylor approximation regarding the IS curve. Bauer and Neuenkirch (2015) were
the first to derive such a framework and found empirical evidence that central banks
indeed take the resulting uncertainty into account. Moreover, their paper provides strong
arguments that linear macroeconomic models are a shortcoming in the scientific literature.
The analysis is preceded by an overview of dynamic stochastic general equilibrium
(DSGE) models with the evolution of the Real business cycle (RBC) theory and the New
Keynesian framework resulting in the New neoclassical synthesis. First, in the analytical
part, demand and supply side (including monopolistic competition and price rigidity)
will yield the NKPC, the New Keynesian contribution to the framework. Second, from
the household’s Euler equation follows the forward-looking IS curve, originating in the
RBC theory. Third, the central bank’s optimization under discretion ends in a (standard)
targeting rule.
The rest of the thesis consists of the augmentation with shocks and uncertainty. After
checking the uniqueness of the equilibrium and the derivation of the equilibrium condition,
initially without uncertainty, the examination is going to be conducted both analytically
via differentials and as numerical simulation in MATLAB. A thorough literature research
discusses possible parameter ranges, with a focus on the persistence parameter. Further-
more, strict inflation targeting is considered. In a very similar way, the quadratic approx-
imation of the IS curve is dealt with. In addition, the case without certainty-equivalence
and persistence is going to be taken into consideration. In a last step, all results will be
compared and interpreted in the context of crisis scenarios, booms or relatively tranquil
times. The comparison of models and settings is of great importance, rather than noting
absolute values.
To keep the framework easily understandable, government, investments, money supply,
and labor markets are omitted. Consequently, neither money holdings nor working hours
(or leisure time) will enter the household’s utility function. I also attach a great deal
of importance to the compelling nature and a clear step-by-step derivation of the NKM.

© Springer Fachmedien Wiesbaden 2017

T. Kranz, Persistent Stochastic Shocks in a New Keynesian Model
with Uncertainty, BestMasters, DOI 10.1007/978-3-658-15639-8_1
That is why, to the best of my knowledge, graduate readers will receive a comprehensive
and comprehensible introduction to this state-of-the-art framework.
The remainder of this thesis is organized as follows: Section 2 reviews the NKM
evolution and derives a basic version. Section 3 expands this with shocks, discusses the
equilibrium condition, and simulates the results. Section 4 adds uncertainty to the model,
simulates, and compares the findings with those in Section 3. Section 5 concludes.

2
2 New Keynesian Model
Years before the components of the New Keynesian framework were developed or used,
there was already a major turning point in macroeconomic modelling, initiated by the
prominent Lucas critique. Lucas (1976) denounced the shortcomings of economic policies
that try to exploit relationships on the basis of (highly aggregated) historical data, such
as the Phillips curve. As a reaction, households and firms would adapt (or even become
forward-looking) and relationships could change or, in the case of the Phillips curve, break
down.
The following historical recapitulation starts in the early 1980’s and describes how the
different NKM’s origins have developed and coalesced since. Subsequently, the theoretical
part will be introduced including the NKPC (i.e., demand side, supply side, and Calvo
pricing), the forward-looking IS curve, and a monetary policy rule under discretion.

2.1 The New Neoclassical Synthesis

The paper by Kydland and Prescott (1982) is considered to be the starting point in RBC
theory. They explained fluctuations in business cycles with a non-time-separable util-
ity function and focussed on intertemporal substitution. Prescott (1986) added shocks
to technology and examined people’s willingness and ability to intertemporally and in-
tratemporally substitute in detail. Generally, the RBC theory established the use of fully
specified DSGE models with characteristic features as the efficiency of business cycles,
the importance of technology shocks (as a source of fluctuations in economic output), and
the limited role of monetary factors (see Galı́ 2015, 2–3).
During the same time, in the 1980’s, models with market imperfections were devel-
oped. Akerlof and Yellen (1985) gave an explanation of why variations in the nominal
supply of money are not neutral in the short run. They emphasized the suboptimal re-
actions to demand shocks through price (and wage) inertia and that it is important to
implement this property in RBC models. Mankiw (1985) discussed the conflict between
modern neoclassical and traditional Keynesian theories. He modeled price inertia with
fix costs (menu costs) and also assumed a monopoly, another market friction the RBC
theory neglected. Blanchard and Kiyotaki (1987) came to the result that monopolis-
tic competition, together with other imperfections, can explain why aggregated demand
affects output.
The first derivation of a NKPC can be found in the paper by Roberts (1995, 979).
Neither the name nor the mathematical methods (i.a. Calvo pricing) have fundamentally
changed since. Early (complete) New Keynesian frameworks were modeled by Yun (1996)

© Springer Fachmedien Wiesbaden 2017

T. Kranz, Persistent Stochastic Shocks in a New Keynesian Model
with Uncertainty, BestMasters, DOI 10.1007/978-3-658-15639-8_2
and Rotemberg and Woodford (1997). The latter explicitly referenced the Lucas critique
and discussed the new era of macroeconomic models. In the same year, Goodfried and
King (1997) proposed four elements typical for the combination of NKMs and RBC theory:
Intertemporal optimization, rational expectations, imperfect competition, and costly price
adjustment. They labeled it New neoclassical synthesis, in reference to the Neoclassical
synthesis, which was developed and popularized in the decades after Keynes’ main work
and combined Keynesian macroeconomics with neoclassical thoughts.
In a next step, Clarida et al (2000) combined these thoughts and made big strides in
the field of monetary policy. They surveyed the post-war United States economy in terms
of forward-looking monetary policy implications. For their baseline model, they referenced
i.a. Yun’s paper and work by Woodford. As soon as DSGE models proved themselves to
be capable of describing and explaining short-run fluctuations, price stickiness, and evalu-
ating monetary policy; central banks adopted this and developed large-scale models. The
most notable representative is the DSGE model by Smets and Wouters (2002), providing
a theoretical foundation for the work done by the ECB.
The widely respected paper by Christiano et al (2005) showed that stickiness in nomi-
nal wages is more important than price sluggishness. By that time, these models included
common features like households’ habit persistence, that describes how past consumption
influences the marginal utility of present consumption. Furthermore, they included Calvo
pricing in both wages and prices, variable capital utilization, costs for capital adjustment,
and a variety of shocks. These models implicate a monetary policy that is aimed to sta-
bilize the economy. Generally, stabilizing inflation is sufficient, since a stabilized output
goes hand in hand with maintaining perfectly stable inflation rates. In the article by
Blanchard and Galı́ (2007), this property was named the “divine coincidence”.
In the process evolving over the past decades, the main purpose was to understand
(i) fluctuations, (ii) rigidities, and (iii) monetary policy, which can affect real output in
the short run. Therefore, the New Keynesian framework provides the theoretical basis for
much of the contemporary macroeconomic modeling.

4
2.2 New Keynesian Phillips Curve

For deriving the NKPC, two optimization problems concerning private households and
firms are employed, leading to aggregated demand and supply. Furthermore, price rigidity
is modeled through the method introduced by Calvo (1983). The time index t is only
used from the Calvo Pricing section onwards, where it is needed to make a distinction
between the different periods.

Demand Side

On the demand side, the representative consumer can choose from a variety of goods Cξ
which results in an average consumption of C. Usually, the CES function is used to model
monopolistic competition1 , one of the two market frictions incorporated into the NKPC:
 1  ε−1
ε
ε−1
C= Cξ ε dξ . (1)
0

Here, ξ ∈ [0, 1] can be viewed as a continuum of firms from 0 to 100%. The exponent is
a measure for the substitutability between the goods Cξ , where ε represents the elasticity of
substitution. A change in ε would result in a different mean for C.2 Bauer and Neuenkirch
(2015, 5) assume ε > 1, constant over time and common amongst all economic subjects.
A Hicksian-like3 optimization by means of the Lagrangian function helps to solve for
the demand curve:
   ε−1
ε 
1 1 ε−1
L (Cξ , λ) = Pξ · Cξ dξ − λ Cξε
dξ −C . (2)
0 0

Since firms have pricing power, the representative consumer takes prices Pξ as given.

Minimizing expenditures Pξ Cξ with the constraint of a certain consumption level C
requires the following first-order conditions4 :

 1  ε−1
1
∂L −1 ε−1
= Pτ − λCτ ε Cξ ε dξ =0 (3.1)
∂Cτ 0
 1  ε−1
ε
∂L ε−1
= Cξ ε dξ −C = 0. (3.2)
∂λ 0

1
Dixit and Stiglitz (1977) developed this approach. However, they used a discrete sum and no integral
but received the same results.   1/q
n
2
Eq.(1) is an integral notation for the generalized mean x̄ = n1 j=1 xqj with q = ε−1
ε .
See Appendix A.1 for a more detailed comparison.
3
Since Hicksian demand will be determined by minimizing expenditures with a utility constraint, the
present optimization problem will transfer to this type using the simple utility function U (C) = C.
4
Note that τ denotes a continuum of derivatives.

5
 Differentiating with respect to λ provides the constraint, Eq.(1). Rearranging5 (3.1)
and defining λ ≡ P 6 as the aggregated price level yields
 ε
P
Cτ = C, (4)
Pτ

the demand7 for good i. Substituting this in Eq.(1) and rearranging8 gives the formula

 1  1−ε
1

P = Pξ1−ε dξ , (5)
0

which describes the aggregated price level9 . The lack of investment and governmental
spendings in this model leads to Yτ = Cτ . Each firms’ production Yτ will be consumed
completely by private households and hence Y = C. Eq.(4) can now be written in the
aggregated demand form
1
Yτ = · Y P ε, (6)
Pτε
that takes the shape of a hyperbola and is, in consequence, very similar to the tradi-
tional downward sloping AD curve.10

Supply Side

Each firm takes the aggregated demand function and the aggregated price level P as
given since any single firm is too small to directly influence other prices or productions.
It chooses its own price Pτ and faces the typical (real) profit maximization problem
 
Pτ Yτ
max − K(Yτ ) , (7)
Pτ ,Yτ P

5
See Appendix A.3 for more details.
6
There are good reasons for this step. When the consumption constraint is relaxed by one unit,
total consumption expenditures (see Galı́ (2015, 53)) will increase to (C + 1)P = CP + P , where P is
the amount by which the optimum will change. This is exactly the information the Lagrange multiplier
λ contains. See Chiang and Wainwright (2005, 353–354) for a detailed proof with λ expressed as a
derivative.
7
It is straightforward to show that ε is in fact the elasticity of substitution. Dividing two demand
functions (for goods a and b) by each other results in (Cb /Ca ) = (Pa /Pb )ε . Since ε is defined as the
percentage change in relative goods by a percentage change in relative prices, this is exactly what the
equation shows. See Chiang and Wainwright (2005, 396–399) for the definition and the case of the CES
production function.
8
See Appendix A.4 for more details.
9
Again, like Eq.(1), this can be examined through the generalized mean, see Appendix A.5.
10
The concise paper of Kyer and Maggs (1992) recapitulates the topic and analytically derives an AD
curve consisting of both the Keynes and the Pigou effect.

6
 where it substracts the real costs K from the real revenue. Using (6) and rearranging
leads to  1−ε  −ε 
Pτ Pτ
max Y −K Y . (8)
Pτ P P

The first-order condition is now straightforward, using the chain rule:

 −ε  −ε−1
∂ Pτ Y Pτ Y
= (1 − ε) · − K  (Yτ ) · (−ε) · = 0. (9)
∂Pτ P P P P

Simplifying and denoting the optimal price with Pτ∗ yields

 −ε  ∗ −ε−1
Pτ∗ Pτ
(ε − 1) = K  (Yτ ) · ε (10.1)
P P
   ∗ −1
ε  Pτ
⇔ 1= K (Yτ ) (10.2)
ε−1 P
 
∗ ε 
⇔ Pτ = K (Yτ ) · P, (10.3)
ε−1

an important result that states that the optimal price Pτ∗ equals the nominal marginal
costs and a mark-up (ε/(ε − 1)) > 1 for all ε > 1. However, perfect substitutes let the
monopolistic structure vanish and show the typical polypolistic result:
 
ε
lim K  (Yτ ) · P = K  (Yτ ) · P = Pτ∗ . (11)
ε→∞ ε−1

Now, with a cost function in real terms of quantities Yτ defined11 as

cvar γ+1
K(Yτ ) = Y + cf ix , (12)
γ+1 τ

where cf ix are the fix costs, cvar is a measure for the variable costs and γ represents
the elasticity of marginal costs, (10.3) becomes a micro-funded AS curve that takes the
form of a power function:  
ε
Pτ∗ = cvar Yτγ · P. (13)
ε−1

Log-linearization. It is convenient to use log-linearized variables instead of level

variables to be able to solve the model analytically. Also, some interpretations of the

11
Bauer and Neuenkirch (2015, 6) showed that an explicit formulation of the cost function is not
necessary. However, to give a better understanding of the mechanics behind the NKPC, an actual
function is used.

7
results, in terms of elasticity and growth rates, become quite useful.12 So both (6) and
(13) will now be approximated through log-linearization around the steady state.13 But
first some preparation is necessary.
Let Z be a state variable that can change over time and Zss its long-term value. When
defining
z ≡ ln Z − ln Zss , (14)

z becomes a good approximation of z, the growth rate around the steady state.14 Fur-
thermore, in the steady state, long-term values for individual variables are by definition
the same as for those on aggregated level, thus Zτ ss = Zss . The state would otherwise in-
clude endogenous forces. And finally, the long-run marginal costs equal the multiplicative
inverse of the firms’ mark-up:15
ε−1
cvar Yssγ = . (15)
ε
An explanation for that would be the long-run version of Eq.(13) and hence Pτ ss = Pss .
Now this can be applied to the previous results. First, Eq.(6), the AD curve will be log-
linearized. Taking logs, expanding with the log long-term values, and using (14) gives

ln Yτ = ln Y + ε(ln P − ln Pτ ) (16.1)
⇔ ln Yτ − ln Y = −ε(ln Pτ − ln P ) (16.2)
⇔ ln Yτ − ln Yss − (ln Y − ln Yss ) = −ε(ln Pτ − ln Pss − (ln P − ln Pss )) (16.3)
⇔ yτ − y = −ε(pτ − p) (16.4)
⇔ yτ = −εpτ + εp + y, (16.5)

a linearized AD curve in terms of growth rates with the slope of −1/ε. A higher
elasticity of substitution would result in a flatter curve, so a change in the firm’s price
growth pτ would have a stronger effect on production growth yτ .

12
See Romer (2012, 207), Judd (1998, 200–202), the paper by Uhlig (1999), and the guide by Zietz
(2008).
13
The approximation becomes more precise with smaller growth rates, that is exactly what the steady
state can offer. Judd (1998, 196–198) provides a goodness of fit comparison.
14
See Appendix A.6 for a detailed calculation regarding z and the actual growth rates z from one
period to the next. Incidentally, there is no need to use z in the basic model. Furthermore, a first-order
Taylor approximation “in reverse”
 shows
 the relationship
  between
 z and z:
z ≈ ln(1 + z) = ln 1 + Z−Z
Zss
ss
= ln 1 + ZZss − 1 = ln ZZss = ln Z − ln Zss ≡ z.
15
Other authors simply define this property, see e.g. Galı́ (2015, 57).

8
 Next, with the use of (15), the AS curve type Eq.(13), can be rewritten in a similar
way:
 
ε
ln Pτ∗ = ln + ln cvar + γ ln Yτ + ln P (17.1)
ε−1
 
ε
⇔ ln Pτ∗ − ln P = ln + ln cvar + γ (ln Yτ − ln Yss + ln Yss ) (17.2)
ε−1
 
ε
⇔ p∗τ − p = γyτ + ln + ln cvar + γ ln Yss (17.3)
ε−1
 
ε−1
⇔ p∗τ − p = γyτ + ln (cvar Yssγ ) − ln (17.4)
ε
 
∗ cvar Yssγ
⇔ pτ − p = γyτ + ln . (17.5)
(ε − 1)/ε
 
=0

The latter expression shows the assumption that the log deviations of marginal costs
from their long-run trend values are linear in the amount of γ. When the firm’s optimized
price growth p∗τ is equal to the aggregated price growth p, then there is no growth in the
firm’s production. Having log-linearized both demand and supply side, Figure 1 sums up.

pτ

y
p+
ε
slo
pe γ
:− pe:
1/ slo
ε

γy
+p
1 + εγ

y εp + y yτ
1 + εγ

Figure 1: Graphical results of households’ and firms’ static optimization.

9
 Finally, inserting (16.5) in (17.5) combines all the results and gives

p∗τ − p = γ(−εp∗τ + εp + y) (18.1)

⇔ p∗τ − p = −γε(p∗τ − p) + γy (18.2)
⇔ (1 + γε)(p∗τ − p) = γy (18.3)
 
γ
⇔ p∗τ − p = y. (18.4)
1 + γε

Recalling that y can be approximated with y, the GDP growth rate around the steady
state, and using αγ ∈ [0, 1[ as a summarizing parameter, Eq.(18.4) yields

p∗τ − p = αγ y, (19)

a description of the steady state output growth rate, depending on price growth and
microeconomic behavior. The next section introduces a non-optimal price setting scheme
which replicates the actual observed economic patterns.16

Calvo Pricing

Nominal rigidities as the second market friction in the basic NKM are implemented
through the assumption that the firms’ infrequent price adjustment follows an exoge-
nous Poisson process17 , where all firms have a constant probability (φ) to be unable to
update their price in each period with φ ∈ [0, 1[ (i.e., φ = 0 in the absence of price rigid-
ity). It is crucial that price-setters do not know how long the nominal price will remain
in place. Only the expected value is known due to probabilities that are all equal and
constant for all firms and periods. This implies a probability of φj for having the same
price in j periods as today and so the average expected duration between price changes
will be 1/(1 − φ).18
From now on, the time index t will be used, as more than one period is being con-
sidered. Simultaniously, the firm index τ is no longer important since it is sufficient to
calculate with a share of firms φ (or 1 − φ). Hence, p∗τ ≡ p∗t and p ≡ pt . When xt is the

16
See the survey by Taylor (1999), that came to abundant evidence. See also Galı́ (2015, 7–8) for a
literature overview.
17
Calvo (1983) originally wrote his article in continuous time. However, using discrete periods im-
mensely helps the clearness and is more realistic with regard to how the central bank actually operates.
Moreover, Calvo (1983, 396–397) shows the equivalence of both approaches. An alternative model of
sticky prices was provided by Rotemberg (1982). See also Ascari and Rossi (2008) for a comparison of
both approaches in the context of the NKPC.
18
See Appendix A.7 for proof.

10
price that firms will set in period t (provided they are able to do so), the following will
apply:

pt = φpt−1 + (1 − φ)x (20.1)

    t
share of sticky prices share of price adjuster
pt − φpt−1
⇔ xt = (20.2)
1−φ
Et pt+1 − φpt
⇒ Et xt+1 = . (20.3)
1−φ

Firms will act on the probability of not being able to adjust prices in future periods.
In consequence, they try to set a price xt that is not necessarily the optimal price p∗t ,
derived in the previous section. Also, in the presence of price rigidities, xt = p∗t generally
holds.
To reveal the mechanics behind the staggered price setting, it is convenient to verbally
treat pt and xt as level variables. Strictly speaking, the firms set price growth paths in
the following optimization problem rather than maximizing a discounted profit as the
difference between revenue and costs19 . This is due to the fact that it is more in line
with the other microfoundations in the model. In the present way, the optimal reset
price, determined by the discounted sum of future profits, is derived through a quadratic
approximation of the per-period deviation from maximum-possible profit with β ∈ [0, 1[,
the discount factor over an infinite planning horizon. Therefore, firms minimize their loss
function, the discounted deviations from p∗t over all t:
 ∞

2
min Et k β j φj xt − p∗t+j . (21)
xt
j=0

The parameter k > 0 enters the loss function multiplicatively and indicates all exoge-
nous factors that will influence the costs of not setting the optimal price in each period.20
The first-order condition is
∞

∂
= Et 2k (βφ)j (xt − p∗t+j ) = 0. (22)
∂xt j=0

Dividing by 2k, using the fact that xt is t-measurable, and expanding the sum gives
∞ ∞
(βφ)j xt − (βφ)j Et p∗t+j = 0. (23)
j=0 j=0
19
See Walsh (2010, 241–242) for the use of level variables in Calvo pricing.
20
Note that it can also come up as an additive term or any other positive monotonic transformation
and does not alter the results.

11
 Excluding xt from the sum, using the formula for an infinite geometric series, and
multiplying by (1 − βφ) gives

∞
xt = (1 − βφ) (βφ)j Et p∗t+j . (24)
j=0

Again, using t-measurability (Et p∗t = p∗t ) and excluding the first summand provides a
sum from j = 1 to infinity that can be substituted in a subsequent step:

∞

xt = (1 − βφ) (βφ) j
Et p∗t+j + p∗t . (25)
j=1

Furthermore, Eq.(24) can be rewritten for t + 1 (since firms optimize in each period),

∞
Et xt+1 = (1 − βφ) (βφ)j−1 Et p∗t+j (26.1)
j=1
∞
⇔ βφEt xt+1 = (1 − βφ) (βφ)j Et p∗t+j , (26.2)
j=1

for eliminating the sum in (25):

xt = βφEt xt+1 + (1 − βφ)p∗t . (27)

Inserting (20.2) and (20.3) leads to the expression

pt − φpt−1 Et pt+1 − φpt

= βφ + (1 − βφ)p∗t (28.1)
1−φ 1−φ
⇔ pt − φpt−1 = βφ(Et pt+1 − φpt ) + (1 − φ)(1 − βφ)p∗t , (28.2)

that only contains parameters and variants of the variable p. Then, with the definition
of (14) and first-order Taylor expansion, the inflation rate π can be expressed through
differences of p:

pt − pt−1 = ln Pt − ln Pss − (ln Pt−1 − ln Pss ) = ln Pt − ln Pt−1

   
Pt Pt − Pt−1
= ln = ln + 1 = ln(πt + 1) ≈ πt . (29)
Pt−1 Pt−1

In the same way, the conditional expectation value for period t + 1 can be expressed
with
Et pt+1 − pt ≈ Et πt+1 . (30)

12
 Since this approximation is sufficiently exact for small values of π, an equality sign
will be used for all following calculations. Now (28.2) can be rearranged to insert approx-
imations (29) and (30):

φ(pt − pt−1 ) = βφ(Et pt+1 − φpt ) + (1 − φ)(1 − βφ)p∗t − (1 − φ)pt (31.1)

(1 − φ)(1 − βφ) ∗ 1 − φ
⇔ πt = βEt πt+1 + pt − pt + β(1 − φ)pt . (31.2)
φ φ

Isolating (p∗t − pt ) and replacing it with the result from last section, Eq.(19), gives

αγ (1 − φ)(1 − βφ)
πt = βEt πt+1 + yt . (32)
φ

Usually21 , in a final step, a summarizing parameter κ > 0 for all parameters, multiplied
with yt , will be defined to end up in the NKPC:

πt = βEt πt+1 + κyt . (33)

Both the expected inflation rate Et πt+1 and the GDP growth rate around the steady
state yt (or output gap) have a positive impact on πt since β, κ > 0. Moreover, the slope
of the NKPC (κ), depends on all four parameters (β, γ, ε, and φ) of this section.22
So far, households’ optimization led to aggregated demand and firms’ profit opti-
mization led to aggregated supply. Price rigidity was modeled via Calvo pricing which
received the NKPC. In the next section, private households optimize their consumption
intertemporally under a budget constraint. This leads to a forward-looking IS curve.

2.3 Forward-Looking IS Curve

The objective is to derive an Euler equation via maximizing utility with a dynamic budget
constraint. Initially, it is not necessary to formulate an explicit utility function. On
the contrary, the general marginal utility gives a better insight into the intertemporal
mechanics. Only one specific assumption will be made, namely not considering money,
working hours or any other possible utility-gainer. It solely relies on consumption and
thus households maximize their intertemporal discounted utility
 ∞

max Et β s−t U (Cs ) , (34)
Ct
s=t
21
See e.g. Galı́ (2015, 63) or Walsh (2010, 338).
22
See Appendix A.8 for a more detailed analysis of κ. Depending on the exact model, the slope of the
NKPC can have a slightly different meaning, e.g. Walsh (2010, 336) with a measure for the firm’s real
marginal costs instead of the output gap.

13
 under the constraint

Ct · Pt + Bt+1 = Wt + (1 + it−1 ) · Bt , (35)

where W is the nominal wage and B the amount of bonds. The latter provides the
link between two periods.23 The Lagrangian brings both together:

∞

L (Ct , Ct+1 , Bt+1 ) = Et β s−t U (Cs ) − λs (Cs Ps + Bs+1 − Ws − (1 + is−1 )Bs ) . (36)
s=t

Here, the control variable is s, while t always designates the starting period. Let W
be exogenous, then the households’ first-order conditions24 are

∂L
= U  (Ct ) − λt Pt =0 (37.1)
∂Ct
∂L
= βEt [U  (Ct+1 )] − λt+1 Et [Pt+1 ] = 0 (37.2)
∂Ct+1
∂L
= −λt + λt+1 (1 + it ) = 0. (37.3)
∂Bt+1

Inserting (37.1) and (37.2) in (37.3) yields

βEt [U  (Ct+1 )] U  (Ct ) Pt

(1 + it ) = ⇔ U  (Ct ) = β · (1 + it ) · Et [U  (Ct+1 )], (38)
Et [Pt+1 ] Pt Et [Pt+1 ]

the Euler equation, revealing the intertemporal relationship of the marignal utility out
of consumption.25 Marginal utility in period t equals the counterpart in t+1, corrected by
discount factor, nominal interest rate, and the ratio of current and expected future price
level. Assuming it rises, marginal utility in t would also rise relative to period t + 1. Given
the diminishing marginal utility property and therefore concavity, consumption will be
higher in the future.26

23
Depending on the definition of the interest rate, the period can vary. Here it has been chosen in a
way so that the interest from period t enters the Euler condition.
24
Differentiating with respect to Ct+1 is possible because of linearity and Fatou’s lemma regarding the
conditional expectation.
25
See in Appendix A.10 for more about the mechanics behind intertemporal optimization by an alter-
native calculation via Dynamic Programming.
26
Note that present consumption could also increase because of the income effect.

14
 One possible formulation27 for such a function is U (Ct ) = (1 − σ)−1 · (Ct1−σ − 1) with
σ > 0 implying 1/σ as the intertemporal elasticity of substitiution (IES).28 Substituting
this in the Euler equation gives

Pt
Ct−σ = β · (1 + it ) · −σ
Et [Ct+1 ], (39)
Et [Pt+1 ]

which can be expressed as

Pt
Yt−σ = β · (1 + it ) · −σ
Et [Yt+1 ], (40)
Et [Pt+1 ]

when recalling the market clearing condition Y = C. The long-run real interest r
enters the equation through β since it equals 1/β − 1.29 Solving for Yt results in the
typical downward sloping IS relationship with real interest and expectations expressed in
level variables or including Et πt+1 :
 1/σ
1 Et [Pt+1 ] σ
Yt = 1/σ
· (1 + r) Et [Yt+1 ]
(1 + it ) Pt
1 σ 1/σ
= · (1 + r)(1 + Et πt+1 )Et [Yt+1 ] . (41)
(1 + it )1/σ

To prepare the approximation, Eq.(40) will be rearranged. Treating t-measurable

variables as constants for the conditional expectation, assuming30 Covt [Pt+1 , Yt+1
σ
] = 0,
and taking logs yields
  σ

Pt+1 Yt+1
ln Et · σ = ln (β(1 + it )) . (42)
Pt Yt

27
In an early study, Pratt (1964, 132–134) examined the different classes of utility functions in the
context of risk aversion. See also Kranz (2013, 6–8) for an alternative subdivision into square root,
logarithm, exponential, and broken rational functions.
28
The first study about Euler equations describing intertemporal private consumption came from Hall
(1978). The Appendix A.11 shows that 1/σ is indeed the intertemporal elasticity of substitution with
respect to (39) and (40).
29
The relation follows from the steady state Euler equation or see e.g. the Ramsey–Cass–Koopmans
model, where in the utility maximizing steady state, the net marginal product of capital equals 1/β − 1
and therefore β = 1/(1 + r). See Galı́ (2015, 132) for a more complex definition of the long-term real
interest rate.
30 σ σ σ
Since Et [Pt+1 ]·Et [Yt+1 ] = Et [Pt+1 ·Yt+1 ]−Covt [Pt+1 , Yt+1 ], the values are potentially overestimated.
However, the assumption at this point (in contrast to Section 4.1) does not alter the results and is only
for clarity. The following linearization eliminates second order moments.

15
 With the help of the exponential function, prices and output can be rewritten in their
log deviations:
   σ

Pt+1 Yt+1
ln Et exp ln · σ = ln Et [exp (ln Pt+1 − ln Pt + σ(ln Yt+1 − ln Yt ))]
Pt Yt
= ln Et [exp (pt+1 − pt + σ(yt+1 − yt ))] . (43)

The next steps involve approximations of both inflation rate and output gap and also a
first-order Taylor expansion of both the exponential and the logarithmic function:

t+1 − y
⇒ ln Et [exp (πt+1 + σ(y t ))] (44.1)
t+1 − y
≈ ln Et [1 + πt+1 + σ(y t+1 − y
t )] = ln (1 + Et [πt+1 + σ(y t )]) (44.2)
t+1 − y
≈ Et [πt+1 + σ(y t+1 − σ y
t )] = Et πt+1 + σEt y t . (44.3)

Steps (44.2) and (44.3) were to bypass Jensen’s inequality31 since the function’s cur-
vature is sufficiently small32 . Together with the right side of Eq.(42), after expressing β
with r, log-linearizing, and rearranging,

1
t+1 −
yt = Et y (it − r − Et πt+1 ) (45)
σ

follows, a dynamic IS curve with the output gap instead of the actual GDP. Referring
to the original graphical IS relation, the curve shifts to the right if the long-term real
interest rate r, the output gap expectations Et y
t+1 or the inflation expectations Et πt+1

rise. However, the slope will rise and the curve becomes flatter if the intertemporal
elasticity of substitution (1/σ) rises.
After a dynamic utility optimization that yielded the Euler equation, the result could
be approximated to an equation dependend on the nominal interest rate it , explaining the
output gap yt . Until now, there are two equations describing the economy, a linearized
Phillips curve and a linearized forward-looking IS curve. Consequently, the equation
system cannot be solved for the three t-measurable variables yt , πt , and it . The next
section adds a third equation, originating from a loss function by the central bank that
links output gap to inflation rate.

31
See Jensen (1906, 190–192) that f (EX) ≥ E[f (X)] holds for concave functions, i.e. the logarithm.
See also Billingsley (1995, 449) that Jensen’s inequality still holds for the conditional expected value.
32
The accuracy is comparable to log-linearization for small growth rates. Moreover, the exactness
increases for larger values because of (ln(x)) → 0 for increasing x. However, resulting values will always
be underestimated.

16
2.4 Targeting Rule under Discretion

The central bank takes NKPC and IS curve as given and wants to optimally set the
interest rate for period t. There are several ways to proceed and before describing the
optimization problem, a short summary introduces the different concepts.
An early discussion on how to model policy decisions can be found in the work by
Kydland and Prescott (1977). Also, Givens (2012, 1) provides an excellent overview:
“In macroeconomic models that embody rational expectations, optimal monetary policies
are separated by a dichotomy known in the literature as commitment and discretion.” He
further writes about discretion: “Changes in the interest rate are [. . . ] the result of period-
by-period reoptimization in which foregoing policy intentions are considered irrelevant for
current decision making (Givens 2012, 1).” In contrast, under commitment, the central
bank has a mandatory inflation target and is able to affect expectations. However, a
discretionary monetary policy is more flexible, since the interest or inflation rate (path)
is not fixed. Drawbacks could be credibility problems that arise if the central bank has
to readjust the interest rate frequently due to a volatile economy or if it sets the interest
rate in big leaps. The latter is rarely seen in practice and the more realistic central bank
behavior is referred to as inertia, “[. . . ] a series of small adjustments in the same direction,
drawn out over a period of months, rather than through an immediate once-and-for-all
response to the new development (Woodford 2003a, 1).”
Because of the far simpler approach, it is more tempting to start with an optimization
unter discretion. Therefore, the central bank’s targeting rule will be derived by minimizing
the discounted loss function over all periods33
 ∞

s−t ∗ 2 2
min Et β (πs − π ) + δ ys , (46)
π, y
s=t

under the constraints (33) and (45), the Phillips curve and IS curve respectively.
Every difference between the inflation rate and the central bank’s target π ∗ results in a
loss. In the first instance, π ∗ = 0 as it does not change the essential findings. Also, every
output gap leads to a loss but is reduced by a weighting factor δ, normally smaller than
one. Squaring ensures that higher deviations yield disproportionately higher losses and
the optimized variables will not vanish in the derivatives. Moreover, it makes the loss

33
The loss function can be derived by a second order approximation of the households’ welfare loss,
first introduced by Rotemberg and Woodford (1999, 54–61). It can also be found in the textbooks by
Galı́ (2015), Walsh (2010), and Woodford (2003b).

17
function symmetrical34 . The Lagrangian has to be differentiated with respect to yt , πt ,
and it , since the central bank sets the nominal interest rate:

∞
L (πt , yt , it ) = Et β s−t πs2 + δ ys 2 − χs (πs − βπs+1 − κys )
s=t

1
−ψs (ys − y
s+1 + (is − r − πs+1 )) . (47)
σ

First-order conditions:

∂L
= 2πt − χt =0 (48.1)
∂πt
∂L
= 2δ yt + χt κ − ψt = 0 (48.2)
∂ yt
∂L ψt
=− = 0. (48.3)
∂it σ

From condition (48.3) follows that ψt = 0, hence the minimized loss will not change if
the IS curve shifts, as the central bank can counteract it one by one through resetting the
nominal interest rate.35 Combining (48.1) and (48.2), the standard targeting rule under
discretion arises:
κ
δ yt = −κπt ⇔ yt = − πt . (49)
δ
Although the optimal interest rate is not explicitly given, all relationships between
the macroeconomic variables are derived. The process is as follows: the nominal interest
rate has an effect on the output gap (IS curve), which in consequence affects the inflation
rate (NKPC). Furthermore, Eq.(49), the “leaning against the wind” condition, implies a
countercyclical monetary policy, that is, to stabilize prices and eventually contract the
economy. The degree of this contraction increases in κ and decreases in δ, the weight on
output stabilization. Former President of the EBC, Jean-Claude Trichet, picked this up
in one of his speeches36 : “The leaning against the wind principle describes a tendency to
cautiously raise interest rates even beyond the level necessary to maintain price stability
over the short to medium term [. . . ].”
Despite these explanations, Eq.(49) could give the impression of the central bank
acting too contractive (i.e. hawkish37 ). Every (positive) deviation from the inflation

34
See Nobay and Peel (2003, 661) for an asymmetric loss function (Linex form) that becomes quadratic
in a special case.
35
Blanchard and Galı́ (2007) call this “divine coincidence,”the lacking trade-off in the basic NKM, and
argue that it comes from the absence of real imperfections. Also, Galı́ (2015, 129) takes only the NKPC
and not the IS curve as a constraint since results will not change.
36
“Asset price bubbles and monetary policy,” 8 June 2005, Singapore.
37
Note that, on the other hand, an expansive monetary policy is referred to as dovish.

18
Exploring the Variety of Random
Documents with Different Content
keeping it always full of water. In the afternoon I made a long
excursion with my Zintáni through the plain, beyond the chapel of
Sidi Bargúb, in order to buy a sheep; but though the flocks were
numerous, none of the shepherds would sell, as pasturage was
abundant and every one had what he wanted.
Foggy weather indicated that rain was approaching; and just in
time Mr. Richardson with his party arrived, and pitched his enormous
lazaretto tent opposite our little encampment. Mr. Reade also had
come from the town, in order to settle, if possible, the
misunderstanding with our servant Mohammed, and see us off. It is
an agreeable duty for me to acknowledge the many services
rendered us during our stay in Tripoli. Our whole party was detained
here the following day by the heavy rains; and Overweg and I were
happy to get hold of the black servant of the ferocious pseudo-sheríf
mentioned by Mr. Richardson, when that troublesome fellow was
sent back to town, as we were much in want of another servant.
We fairly set out on our exhibition. The country became more
diversified as we approached the defile formed by the Bátes and
Smaera, two advanced posts of the mountain-chain, while the varied
forms of the latter, in high cones and deep abrupt valleys, formed an
interesting background. But the country hereabouts is cultivated with
less care than Wady Mejenín; and the ground being more stony,
presents of course more obstacles than the latter, while both districts
are inhabited by the same tribes, viz. the Urgáat and the Akára.
Even here, however, in the circle formed by the surrounding heights,
was a fine extent of plain covered with corn-fields. Just at the
entrance of the pass there is a well, where the road divides; and
after a little consultation, we took the western branch, as our people
feared that on the eastern we should not find water before night.
Changing, therefore, our direction, we seemed awhile to keep off
entirely from the mountain-range till we reached the wide but very
rugged and rocky Wady Haera, which it was our object to reach at
this spot, in order to fill our water-skins from the pools formed by the
rains. The wady, indeed, looked as if it sometimes bore in its floods a
powerful body of water; and a considerable dyke had been
constructed in the early times of the Arabs, extending for two
hundred paces from the wady eastward; but it has fallen to ruin, and
the path leads now through the breach.
Resuming our march, after a good deal of delay, we turned sharp
off towards the mountains, and at an early hour encamped on a very
pleasant spot adorned with numerous sidr-trees (Rhamnus Nabeca);
but instead of enjoying it in quiet, Overweg and I felt disposed to
direct our steps towards a hill called Fulíje, about half an hour’s walk
eastward, which promised to be a convenient point for obtaining
correct angles of the prominent features of the chain, and proved to
be so in reality. Having executed this task, therefore, we returned to
our companions well satisfied, and spent the evening in the
comfortable tent of Mr. Warrington. We had now reached the slope of
the chain, where some of our people supposed that the boat would
cause difficulties; but it could not well do so after being cut into
quarters, which fitted to the sides of the camels rather better than the
large quadrangular boxes. The most troublesome parts were the
long oars and poles, which caused the camel much exhaustion and
fatigue, by constantly swaying backwards and forwards.
The ground, soon after we had started the next morning, became
stony, and, at three miles distance, very rugged and intersected by a
number of dry watercourses. The landscape was enlivened not only
by our own caravan, composed of so many heterogeneous
elements, but also by some other parties who happened to be
coming down the slope: first the Kaimakám of the Jebel, then a slave
caravan, consisting of about sixty of these poor creatures, of whom
the younger, at least, seemed to take a cheerful interest in the varied
features of the country. The Wady Bú Ghelán, where the ascent
commences, is here and there adorned with clusters of date-trees. In
about an hour the first camels of our party reached the terrace of
Beni ʿAbbás; and till the whole had accomplished the ascent, I had
leisure to dismount from my tractable Bú-saefi, and to sit down
quietly under a fine olive-tree near the chapel of the Merábet Sámes,
watching them as they came up one by one, and cheered by the
conviction that the expedition was at length in full train. The country
was here hilly, and the path very often narrow and deeply cut in the
marly soil. Further on, Overweg and I, together with our shoush,
turned off a little to the right from the great caravan-road, and,
passing through fine corn-fields interspersed with flowers of different
kinds, reached the village Gwásem, lying at a short distance from the
eastern foot of Mount Tekút, where we were treated with sourmilk by
a friend of our companions. When we had overtaken our caravan, I
found time to pay a visit to the Roman sepulchre, and ascertained
that the base measured 24 ft. in every direction, the principal body of
the monument, containing the sepulchral chamber, having fallen in
entirely. From this point we began to ascend the second terrace, and
reached the level of the plateau at two o’clock in the afternoon. The
country had now a much more interesting appearance than when I
was here two months before, being at present all covered with green
corn. Having started in the direction of the castle, we descended, a
little before reaching it, along the shelving grounds towards Wady
Rummána, and encamped on the spot where the troops usually
bivouac.
Here we remained the following day, when, in order to settle
formally the demands of our camel-drivers, we had all our things
accurately weighed by the officials of the castle. The little market did
not grow busy till ten o’clock. The chief articles for sale were three
head of cattle, one camel, some sheep and goats, a few water-skins,
some barley, a few eggs, and sandals; but at noon it was moderately
thronged. In the afternoon we paid a visit to several subterranean
dwellings, but were disappointed in not getting access into an
entirely new structure of this kind, formed of a much harder sort of
clay. Our cheerful friend Mr. Warrington, in order to treat our party
before he separated from it for a length of time which nobody could
foresee, got an immense bowl of kuskus prepared, seasoned in the
most savoury manner; and our whole party long indulged in the
remembrance of this delicate dish as a luxury beyond reach. The site
of our encampment was most pleasant: below us the wady, rich with
varied vegetation; while towards the north the Tekút, with its regularly
shaped crater towering proudly over the lower eminences around,
formed a most interesting object. Though busy at an early hour, we
did not get off till late; for many things were still to be settled here.
We separated from Mr. Warrington: and of the three travellers I was
the only one whom he was ever to see again.
 Our path was at first very winding, as we had to turn round the
deep indentation of the Wady Rummána, after which it took a
straighter course, passing through several villages, with their
respective olive-groves, till we reached Bu Sriyán, where the
cultivation of the olive-tree ceased entirely for some distance, and
the country became more open. Here we made another considerable
deviation from our southerly direction, and followed a wide valley
with much cultivated ground. Having reached the village Sémsa,
situated upon an eminence to our right, we turned off eastward into a
very pleasant ravine with an olive-grove, and then began the steep
ascent towards the height Kuléba, which forms the passage over this
southern crest of the plateau. While the camels in long rows moved
slowly onwards, with their heavy loads, on the narrow and steep
rocky path, I, allowing my camel to follow the rest, ascended directly
to the village, which is situated round the eastern slope, and is still
tolerably well inhabited, although many a house has fallen to ruin; for
it has a considerable extent of territory; and owing to its situation as
the southernmost point of Ghurián, the inhabitants are the natural
carriers and agents between the northern districts and the desert. On
the highest crest, commanding the village, there was formerly a
castle; but it has been destroyed by the Turks.
Having descended a little into the barren valley, we encamped, at
two o’clock in the afternoon, on the slope of the western hills, near
the last scanty olive-trees, and not far from the well, from which we
intended to take a sufficient supply of water to last us till we reached
Mizda. While our people, therefore, were busy watering the camels
and filling our water-skins, Overweg and I, accompanied by two of
the inhabitants of the village, who had followed us, ascended a
conspicuous mount, Jebel Toëshe, the highest in the neighbourhood,
on the top of which a village is said to have existed in former times.
We took several angles; but there is no very high point about Mizda
which could serve as a landmark in that direction. The country
through which we were marching, along irregular valleys, mostly of
limestone formation, exhibited scattered patches of corn for about
the first three miles, after which, almost every sign of cultivation
suddenly ceased, and the “Twél el Khamér,” stretching from north-
west to south-east, about two miles distant on the right, formed, as it
were, the northern boundary of the naked soil. On its slope a few
trees of the kind called radúk by the Arabs were seen from the
distance. We then entered desolate stony valleys, famous for the
bloody skirmishes which are said to have once taken place there
between the Urfílla and the Welád Bu Séf, in the time of ʿAbd el Jelíl.
Refreshing, therefore, was the aspect of Wady Ranne, which,
extending from east to south-west, was overgrown with green
herbage, and had two wells.
A little beyond, near the hill, or rather slope, called Shʿabet el
Kadím, the latter part of which name seems, indeed, to have some
reference to antiquity, we found the first Roman milestone, with the
inscription now effaced: but further on, Mr. Overweg, who went on
foot and was far behind the main body of our caravan, succeeded in
discovering some milestones with inscriptions, which he regretted
very much not being able to show me. Hereabouts commences the
region of the batúm-tree, which, with the fresh green of its foliage,
contributes a good deal to enliven and adorn some favoured spots of
this sterile, gravelly tract. To the left of our path were some
remarkable basaltic cones, starting up from the calcareous ridge.
The ground was strewn with numerous flint-stones. About four
o’clock p.m. I went to look at a curious quadrangular and regularly
hewn stone, three feet in breadth and length, but only eight inches
thick, which was standing upright at some distance from the caravan.
It was evidently meant to face the west; but no trace of an inscription
was to be seen. About a mile further on we encamped at the foot of
the western chain, which rose to a height of about three hundred
feet, and formed a narrow cleft with the eastern chain, which at this
point closes upon it. In this corner (which collects the humidity of two
valleys), besides several batúm-trees, a little corn had been sown.
Panthers are said to be numerous in this region.
The next day we directed our march towards the pass, crossing
the dry beds of several small torrents, and a broader channel
bordered by plenty of batúm-trees. After an hour’s march, we had
reached the summit of the pass, which now began to widen, the
heights receding on each side, and a more distant range bounding
the view. We found in the holes of the rocky bottom of Wady
Mezummíta, which we crossed about half-past eight, several pools
of rain-water, affording us a most refreshing drink; but it was quite an
extra treat, owing to recent heavy rains which had fallen here, for in
general the traveller cannot rely on finding water in this place. The
ground becoming very stony and rugged, our progress was
excessively slow—not above half an English geographical mile in
seventeen minutes. The hills on our right displayed to the view
regular layers of sandstone. Another long defile followed, which at
length brought us to a plain called Wady Lilla, encompassed by hills,
and offering several traces of former cultivation, while other traces,
further on, bore testimony to the industry of the Romans. A small
herd of goats, and the barking of a dog, showed that even at present
the country is not wholly deserted. In our immediate neighbourhood
it even became more than usually enlivened by the passage of a
slave-caravan, with twenty-five camels and about sixty slaves,
mostly females.
After having passed a small defile, we at length emerged into the
north-west branch of the valley of Mizda, called here Wady Udé-
Sheráb, the channel of which is lined with a considerable number of
batúm-trees. Crossing the stony bottom of this plain, after a stretch
of three miles more we reached the western end of the oasis of
Mizda, which, though my fancy had given it a greater extent, filled
me with joy at the sight of the fine fields of barley, now approaching
maturity—the crop, owing to the regular irrigation, being remarkably
uniform—while the grove of date-trees encompassed the whole
picture with a striking and interesting frame. So we proceeded,
passing between the two entirely separated quarters, or villages,
distinguished as the upper, “el fók,” and the lower, “el utah,” and
encamped on the sandy open space a little beyond the lower village,
near a well which formerly had irrigated a garden. People going to
Tripoli encamp at the other end of the oasis, as was done by a
caravan of Ghadamsi people with slaves from Fezzán, on the
following day.
Mizda, most probably identical with the eastern “Musti kome” of
Ptolemy, appears to have been an ancient settlement of the
indigenous inhabitants of North Africa, the Berbers, and more
particularly of a family or tribe of them called “Kuntarár,” who even at
present, though greatly intermixed with Arabs, have not entirely
forgotten their Berber idiom. The oasis lies in the upper part of Wady
Sofejín, or rather a branch of it, stretching out from south-west to
north-east, which has in some parts a great breadth. The natural
advantage, or productive principle, of the locality seems to lie in the
circumstance that the humidity carried down by the Wady Sheráb is
here arrested by a hill, and absorbed by the clayey soil. This hill is of
a lengthened form, and consists entirely of gypsum.

The wells have little depth, and the water is drawn to the surface
by means of oxen; but there being at present only three specimens
of this precious animal in the place, the wells are far from being
made use of to the extent which is practicable and has been once
practised, as may be concluded from the pillars which extend to a
considerable distance on the plain. The town, as I said, consists of
two distinct quarters or villages, of which the western one, situated at
the eastern foot of the hill, is by far the larger; it is built exactly in the
character of the ksúr of the Algerian Sahara, with high round towers
decreasing a little in width towards the upper part, and furnished with
several rows of loopholes. The wall, purposely built with a great
many salient and retiring angles, is in a state of decay, and many of
the houses are in ruins; but the village can still boast a hundred full-
grown men able to bear arms. The chief of this village always
resides in it, while that of the other generally lives at some distance
under tents. The circumference of the village, together with the palm-
grove attached to its eastern side, and consisting of about two
hundred trees, is 2,260 paces.
The lower or south-eastern village, the circumference of which is
600 paces, is separated from the former by an interval of about 400
paces, and has at present no palm-grove, all the gardens having
been destroyed or ruined by neglect, and only twenty or thirty palm-
trees now remaining scattered about the place. About 100 paces
further down the declivity of the valley, is a group of three small
gardens surrounded by a wall, but in bad condition; and at about the
same distance beyond, another in the same state. The only
advantage peculiar to this quarter is that of a large “zawíya,” the
principal articles in the inventory of which are eight holy doves. But
this also has now become but an imaginary advantage, as,
according to its learned keeper’s doleful complaints, it is very rarely
visited. In this as well as in the other quarter, all the houses are built
of gypsum. As Mizda is a very remarkable feature in the country, I
thought it worth while to make a particular sketch of the oasis also
from this side.
 This oasis is very diminutive; but two caravan-routes, one from
Murzuk and one from Ghadámes, join at this point. The inhabitants
are of a mild disposition, and enjoy the fame of strict honesty.
Everything is here considered as secure; and the camels which
cannot find food in the neighbourhood, are driven into the green
valley at four or five minutes’ distance, and left there without a
guardian. I make these statements advisedly, as reflections of a
different kind have been made on their character. The people seem
to suffer much from sore eyes. When we asked them about the most
remarkable features of the road before us, they spoke of a high
mount, Teránsa, which, however, we did not afterwards recognize.
 In the afternoon I made an excursion with Overweg to Jebel
Durmán, situated at the distance of a mile and a half south-east. It is
rather a spur of the plateau jutting out into the broad valley, and, with
its steep, precipitous, and washed walls, nearly detached and
extremely narrow as it is—a mere neck of rock—looks much like a
castle. Upon the middle of its steep side is a small zawíya belonging
to the Zintán. The prospect from this steep and almost insulated pile
could not, of course, be very extensive, as the mount itself is on the
general level of the plateau; but we obtained a fine view over the sea
of heights surrounding the broad valley and the several tributaries of
which it is formed. Night was setting in, and we returned to our tent.
 Having heard our Zintáni make frequent mention of an ancient
castle with numerous sculptures, and situated at no great distance, I
resolved to visit it, and set out tolerably early in the morning of the
9th of April, accompanied by the Arab and one of our shoushes. We
had first to send for one of our camels, which was grazing at about
three miles’ distance, in the sandy bottom of the wady south-east
from our encampment. It was only on this occasion that I became
aware of the exact nature of the valley of Mizda, and its relation to
the Wady Sófejín; for we did not reach this latter wady until we had
traversed the whole breadth of the sandy plain, and crossed a
mountain-spur along a defile called Khurmet bu Mátek, at the
distance of at least eight miles from our encampment. This is the
famous valley mentioned, in the eleventh century, by the celebrated
Andalusian geographer El Bekri. Figs and olive-trees adorn its upper
part, which is said to stretch out as far as Erhebát, a district one day
and a half beyond Zintán; barley is cultivated in its middle course,
while wheat, from which the favourite dish ʿajín is made, is grown
chiefly in its lower part near Tawárgha. The valley seems worthy of
better fortune than that to which it is reduced at present; for when we
marched along it, where it ran S. 20 W. to N. 20 E., we passed ruins
of buildings and water-channels, while the soil exhibited evident
traces of former cultivation. I listened with interest to the Zintáni, who
told me that the valley produced an excellent kind of barley, and that
the Kuntarár, as well as the people of Zintán, his countrymen, and
the Welád Bu-Séf, vied with each other in cultivating it, and, in
former times at least, had often engaged in bloody contests for the
proprietorship of the ground. When I expressed my surprise at his
joining the name of his countrymen with those of the other tribes
hereabouts, he gave me the interesting information that the Zintán
had been the first and most powerful of all the tribes in this quarter
before the time of the Turks, and held all this country in a state of
subjection. Since then their political power and influence had been
annihilated, but they had obtained by other means right of
possession in Mizda as well as in Gharíya, and still further, in the
very heart of Fezzán, by lending the people money to buy corn, or
else corn in kind, and had in this way obtained the proprietorship of a
great number of the date-trees, which were cultivated and taken care
of by the inhabitants for a share of the produce. Formerly the people
of Zintán were in possession of a large castle, where they stored up
their provisions; but since the time of the Turkish dominion, their
custom has been to bring home the fruits of their harvests only as
they want them. In Wady Sháti we were to meet a caravan of these
enterprising people.
 While engaged in this kind of conversation we entered a smaller
lateral valley of Wady Sófejín, and reached the foot of a projecting
hill on its western side, which is crowned with a castle. Here it was
that I was to find marvellous ancient sculptures and drawings; but I
soon perceived that it would be as well not to cherish any high
expectations. The castle, as it now stands, is evidently an Arab
edifice of an early period, built of common stones hewn with some
regularity, and set in horizontal layers, but not all of the same
thickness. It forms almost a regular square, and contains several
vaulted rooms, all arranged with a certain degree of symmetry and
regularity. But while we pronounce the main building to be Arab, the
gateway appears to be evidently of Roman workmanship, and must
have belonged to some older edifice which the Arab chieftain who
built this castle probably found in the place,—a conjecture which
seems to be confirmed by several ornamental fragments lying about.
It is a pity that we know so little of the domestic history of these
countries during the period of the Arab dynasties, though a step in
advance has been made by the complete publication of Ebn
Khaldún’s history; else we should regard with more interest these
relics of their days of petty independence. This castle, as well as
another, the description of which I shall subjoin here, though it was
visited some days later, is called after a man named Khafáji ʿAámer,
who is said to have been a powerful chief of great authority in Tunis
no less than in Tarábolus (Tripoli).
 The other ruin, related to this one as well by name as by the style
of its workmanship—but in many respects more interesting, having
been evidently once a place of Christian worship—stands on a
narrow and detached neck of rock in the Shʿabet Um el Kharáb, and,
from its whole plan, appears to have been originally and principally a
church about forty-three feet square, sufficiently large for a small
congregation, and with more art and comfort than one can easily
suppose a Christian community in these quarters ever to have
possessed. Hence greater interest attaches to this building than it
would otherwise deserve. It closes with a plain apsis, in which there
are two openings or doorways leading into an open room stretching
behind it and the side-naves, and is divided into three naves, the
middle one of which is eight paces, and the lateral ones six and a
half wide. The naves are divided from one another by columns with
differently ornamented capitals supporting arches, all in the so-called
round style of architecture. I made purposely a sketch of two
different capitals, in order to show their designs; and I think they are
very characteristic. But it is curious to observe that the walls also
appear to have been originally painted on stucco, though at present
but a small piece of it remains near the corner; hence I conclude that
the date of the painting was later than that of the erection of the
church.
 The front of the building has suffered in some degree from the
depredations of the Arabs, who are said to have carried away a
great many sculptures from this place—as much, indeed, a man
from Mizda would have made me believe, as fifty-five camel-loads.
However exaggerated this statement may be, it is evident that the
whole layer over the entrance was originally covered with
ornamental slabs, while now only two remain to the left of the
doorway; and these, though in the same style of sculpture as the
capitals, would rather seem to have been taken from another edifice.
There are many debatable points involved in the consideration of this
building. The first fact clearly shown is the existence of a Christian
community or a monastery in these remote valleys, as late as the
twelfth century at least, under the protection of a powerful chief; and
this is not at all improbable, as we know that Mohammed expressly
ordered that zealous priests and monks should be spared, and as
we find so many monasteries in several other Mohammedan
countries. That it was not merely a church, but a monastery, seems
plainly indicated by the division into apartments or cells, which is still
clearly to be seen in the upper story. Attached to the north side of the
church was a wing containing several simple apartments, as the
ground-plan shows; and on the south corner of the narrow ridge is a
small separate tower with two compartments. Near this ruin there is
another, which I did not visit, called Ksaer Labayed mtʿa Derayer,
while a third, called Ksaer el Haemer, has been destroyed.
 CHAPTER V.
SCULPTURES AND ROMAN REMAINS IN THE
DESERT—GHARÍYA.

We lost the best part of the morning, our men not being able to
find their camels, which had roamed over the whole wady. Our road
was almost the same as that by which I had returned the previous
day; and we encamped in the Wady Sófejín, on a spot free from
bushes. From this place, accompanied by the Zintáni, I visited, the
next morning, the castle or convent in Shʿabet Um el kharáb, which I
have described, and thence struck across the stony plateau in order
to overtake our caravan. It was a desolate level, rarely adorned with
humble herb or flower; and we hastened our steps to reach our
companions. Here I heard, from the Zintáni, that his father came
every year about this season, with his flocks, to the valleys east of
our road, and that he would certainly be there this year also. He
invited me to go thither with him, and to indulge in milk to the extent
of my wishes; as for myself, I declined, but allowed him to go, on
condition that he would return to us as soon as possible.
Even after we had overtaken the caravan, the country continued in
general very bare; but we passed some valleys affording a good deal
of herbage or adorned with some fine batúm-trees. About five o’clock
p.m. we encamped in Wady Talha, not far from a Roman castle or
tower on a hill to our left. On visiting the ruin, I found it built of rough
stones without cement, being about twenty feet square in the interior,
with rounded corners, and with only one narrow gate, towards the
east. But this was not the only remnant of antiquity in the
neighbourhood, for in front of us, on the plateau, there appeared
something like a tower of greater elevation; and proceeding early the
next morning, when our people had only begun to load, to examine
it, I found it to be a Roman sepulchre, originally consisting apparently
of three stories; but of these only the base and the first story remain,
while the stones belonging to the upper one are now scattered on
the ground, and show that it was ornamented with small Corinthian
columns at the corners. Even in the most desolate spot, everything
left by the Romans has a peculiar finish. The first story, being all that
at present remains, measures 5 ft. 4 in. on the east and west, and 5
ft. 9 in. on the north and south sides. Not far from this sepulchre are
the ruins of another one, of which, however, nothing but the base
remains, if, indeed, it was ever completed. By the time my drawing
was finished, the caravan had come up.

I then passed several detached cones, the steep precipitous sides

of which, formed by the breaking away of the strata, looked like so
many castles, and traversing Wady Marsíd, reached the camels.
They marched to-day at a very good rate, the quickest we had as yet
observed in ordinary travelling—namely, half a mile in twelve
minutes, making a little less than two and a half miles an hour; but
we afterwards found that this had now become our usual rate,
whereas before reaching Mizda we had scarcely ever exceeded two
miles an hour. The load of the camels, of course, had been heavier
in the beginning; but this can hardly be the only reason of the
difference. The greater dreariness of the country, and the impulse of
our camel-drivers and their beasts to get to their homes, must be
taken into account. I must here observe that Overweg and I
measured our rate repeatedly, with a chain provided by Government,
although it was a very fatiguing labour, and injurious to our dignity in
the eyes of our people. Gradually the day grew very uncomfortable,
a hot west wind driving the sand into our faces, and totally obscuring
the sky. Keeping along the Wady Téroth, sometimes more than a
mile wide, we had on our left a broad mount, rising first with a
gradual ascent, but in its upper part forming a steep and lofty wall
called el Khaddamíyeh. Here, too, according to the information of my
faithful Arab, there is said to be a Roman sepulchre. Having passed
a small defile, and crossed another valley, we had other Roman
ruins on our right, a castle as it seemed, and near it something like a
sepulchre; but the sand-storm hardly allowed us to look, still less to
go in that direction.
At three o’clock in the afternoon we turned off to the west into
Wady Tagíje, and encamped near the bed of a torrent eight feet
deep, which amply testifies that, at times, a considerable stream is
formed here, a fact confirmed by the fresh and luxuriant herbage
springing up in many parts of the valley among thick bushes and
brushwood. Nor was it quite desolate even now; for the flocks of the
Welád Bu-Séf were seen, and their tents were said to be not far off.
The upper part of the valley is called el Khúrub.
This hot day proved a dies ater to my Arab, who had gone to visit
his family. Having brought his old father with him, together with a
goat, as a present, and a skin of milk, he unluckily arrived too late in
the morning at our last night’s encampment. He then sent his father
back with the goat, and began to follow us in the hope of soon
overtaking the caravan; but he was obliged to march the whole
intensely hot day without water, and he could not drink the milk in the
skin, which became quite hot, so that he suffered greatly, and arrived
in a very exhausted state.
The fine herbage procured us a whole day’s rest, as the camel-
drivers were in no haste to bring up their camels. Not knowing this,
but yet convinced that we should not start at an early hour, as the
well was at some distance, and following the information received
from the Zintáni, who was himself too lame to accompany me, I had
taken my gun and pistols at an early hour in the morning, and gone
in the direction of the valley to look after a monument. After nearly
two hours’ march I distinguished something like a high pillar, and,
proceeding straight towards it, found it to be one of the richest
specimens of this kind of monument bequeathed to us by antiquity,
and an indisputable proof that these regions, now so poor, must
have then supported a population sufficiently advanced in taste and
feeling to admire works of a refined character.
The monument rises, upon a base
of three steps and in three stories,
nearly to a height of forty-eight feet.
The base contains a sepulchral
chamber 4 ft. 10⅓ in. long, and 4 ft. ⅛
in. broad, with three niches, one on
the north, and two on the east side.
This side was the principal face of the
monument, forming its most
ornamented part. The first story
measures at its base on the east and
west sides 5 ft 5⅜ in., and on the
north and south sides 4 ft. 10⅛ in.: it
consists of six layers of stones, on the
lowest of which is represented a pair
of wild animals, probably panthers,
with their fore legs or paws resting
upon a sepulchral urn, as if they were
watching it; on the next layer above is
seen the handsome bust of a young
female; two layers intervene without
sculpture; and the fifth is ornamented
on all the four sides with hunting-
scenes. The frieze on every side is
formed by four rosettes; but that on the north side has some
additional decoration, the second rosette on that side, from the east,
exhibiting a group of centaurs, and the fourth a cock. Upon this part
of the frieze is a garland of clusters of grapes; then follows the
moulding.
In the second story the third layer forms the sill and lower part of a
false door very richly ornamented, and on the fifth layer a pair of
genii hold a coronal over the door of the sepulchre, a representation
which seems to intimate Christian ideas. Above it a niche contains
the busts of a man and his wife; but on the north side an elderly
woman occupies a niche with her bust, probably in her character as
proprietress of the single sepulchral niche of the tomb below. Above
is an ornament with two bunches of grapes; and then follows the
frieze, of the common Ionic order. The moulding is surmounted by a
pyramidal roof about twelve feet high, which has lost its summit;
otherwise the whole monument, with the exception of the sepulchral
chamber, which has been broken up in search of treasures, is in the
best state of preservation, notwithstanding its very slender
proportions,—a circumstance very remarkable, after a lapse of at
least more than sixteen centuries. No wonder that the natives of
these regions now regard these tall sepulchral monuments, so
strange at present in this land of desolation, as pagan idols, and call
them “sanem;” for I myself, when alone in front of the monument in
this wide, solitary valley, and under the shadow of the deep,
precipitous side of a plateau adjoining the Khaddamíye on the east,
felt impressed by it with a certain degree of awe and veneration.
My sketch being finished, I was still attracted to a greater distance
up the valley by something which seemed at first to be another
monument; but it was only a mark fixed by the Arabs, and served but
to lengthen my march back, which was more slow, as the heat had
set in. But I was well satisfied with my morning’s work; and my
companions were greatly astonished when they saw the sketch. In
the afternoon I made with Overweg another excursion in the
opposite direction, when after an hour’s march we ascended a
height and obtained a most interesting view over this singular tract,
which seems to be the fragmentary border of a plateau torn and
severed by ravines and precipices, so that only wall-like cliffs, rising
like so many islands out of a sea of desolation, indicate its height. A
high craggy ridge towards the west, with precipitous pinnacled walls,
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com