Macroeconomic risks of Mongolia and ways to mitigate them

Abstract

The paper explores some of the most prominent macroeconomic risks Mongolia is facing at this stage of her development and provides recommendations to reduce them. These have been classified into four main themes: Financial market risks, portfolio risk of Mongolian export basket, macroeconomic mismanagement, and institutional deterioration. The most important of these is argued to be the possible deterioration in the institutional quality. Since the point resources are easy to appropriate various interest groups have incentives to distort the institutions so that extracting the rent is easier. Fighting corruption and creating a stable legal environment where the rule of law and the property rights are respected should be high on the agenda.

Full text

Munich Personal RePEc Archive
Macroeconomic risks of Mongolia and
ways to mitigate them
Bataa, Erdenebat
National University of Mongolia
30 June 2012
Online at https://mpra.ub.uni-muenchen.de/72386/
MPRA Paper No. 72386, posted 06 Jul 2016 07:04 UTC

28
MACROECONOMIC RISKS OF MONGOLIA AND
WAYS TO MITIGATE THEM
Erdenebat Bataa1
The First Draft was submitted on 30 June 2012
The Last Draft was submitted on 25 June 2013
Abstract
The paper explores some of the most prominent macroeconomic risks Mongolia is facing at
this stage of her development and provides recommendations to reduce them. These have been
classified into four main themes: Financial market risks, portfolio risk of Mongolian export
basket, macroeconomic mismanagement, and institutional deterioration. The most important of
these is argued to be the possible deterioration in the institutional quality. Since the point
resources are easy to appropriate various interest groups have incentives to distort the
institutions so that extracting the rent is easier. Fighting corruption and creating a stable legal
environment where the rule of law and the property rights are respected should be high on the
agenda.
Keywords: macroeconomic risk, Mongolia, overcrowding, debt overhang, financial risk,
export portfolio diversification, pro-cyclicality.
1 I would like to thank the anonymous reviewer for providing me with constructive comments and suggestions.

Macroeconomic risks of Mongolia and ways to mitigate them
29
1. Introduction
Macroeconomic risks of Mongolia and ways to mitigate them are extremely important and
timely topic. Detailed analysis perhaps should be discussed within a framework of
geographical vulnerabilities such as limited access to international markets, harsh climates and
structural weaknesses of small population and expensive to maintain infrastructural and
administrative units. Moreover, policy maker’s track record of implementing prudent
macroeconomic policies along with the existing policy framework and people’s level of coping
mechanism when a crisis occurs are also important for such a study. The number of risks one
could look at is also potentially very large. The IMF’s Early Warning Exercise (EWE, 2010)
tracks five sectoral and market vulnerabilities such as external sector risks, fiscal risks,
corporate sector risks, asset price, market valuation, bubble spotting and financial markets that
each has at least two sub-divisions, totaling 22 individual risks for a country. However this
paper prefers depth over coverage and touches on several issues that I believe have not been
addressed elsewhere2.
The paper starts with some recent developmental trends in Mongolia that one needs to take into
account in discussing the risks. The main feature is that Mongolian economy is becoming
heavily dependent on mining sector and here I provide some international historical
developmental experiences of such economies.
Discussions afterwards evolve around four main themes. First I discuss financial market risks.
These include a fast credit expansion in the presence of extremely high interest rates,
government’s extensive involvement in the mortgage market and shrinking intermediation rate
and debt overhang.
The second theme is that of the developments on international commodity markets and their
implications for Mongolian export basket. As the export is dominated by coal whose price is
known to fluctuate wildly there is an elevated risk for the overall economy. It’s therefore vital
to pay significant attention to the accumulation of a buffer reserve that will protect the
economy from international market fluctuations.
Then there are the risks induced by the direct involvement of Mongolian governments in the
economy. It’s often argued that the fiscal policy is pro-cyclical and monetary policy is counter-
cyclical in Mongolia. Since what matters most for the economy is the confluence of these two
policies an accurate measurement of their stances is necessary for successful policy
coordination. The issue of exchange rate sluggishness is also explored here.
The last but not the least theme is concerned with domestic economic prospects that tend to
characterize capital intensive mining economies. It starts with the possible risk of institutional
deterioration. The quality of institutions determines the quality of economic policies but the
presence of high rents can undermine efforts to institution building and/or turn them into
”extractive” mode rather than ”developmental”. Then I move on to a possibility of higher
2 For deeper understanding of other relevant macroeconomic risks of Mongolia the reader is advised also to
consult IMF’s EWE for Mongolia, Batdelger’s (2012) risk report of Mongolia, Gan-Ochir’s (2012, in Mongolian)
Mongolia’s vulnerability assessment conducted in line with Seth and Ragab (2012) and Bataa’s (2012b,c)
possible oil shock impacts for Mongolia.

ERI Discussion Paper Series No. 2
30
unemployment. As the economy progresses from agriculture based economy, which is labor
intensive, into capital intensive mineral economy, the natural rate of unemployment is likely to
increase. The prospect of industrialization is restricted by the Dutch disease and thus there is
resultant loss of competitiveness. This highlights the need for high-investment and high-return
educational system. To view it from a different angle Mongolia is moving from a relatively
diffuse agricultural economy towards a point-resource based economy. As a consequence more
income inequality is expected in the future. This is often followed by socioeconomic instability
and this is the next risk discussed. It also re-iterates the need for better education that allows
the poor to benefit from the economic boom and abolishment of some existing government
regulations that may add to the tension.

Macroeconomic risks of Mongolia and ways to mitigate them
31
2. Changes in industrial structure and macroeconomic risks
Mongolia possesses a large amount of minerals, such as coal, copper, and gold, among others,
and is emerging as a large producer and exporter of fuel. It overtook Australia in 2011 as the
main coal exporter to China (Credit Suisse, 2012). Mining is emerging as the most vibrant
economic sector in terms of GDP contribution (Figure 1), export earnings (Figure 3) and tax
bills (Figure 5).
Real GDP is more than doubled in 2011 compared to its 1995 level, implying an annual
average growth rate of 6%. The mining and the associated boom in net taxes on products and
non-tradable sectors have been the engine of this growth, rising on average by 5.5%, 23% and
7.3% a year respectively. In contrast, agriculture rose only by 0.8% a year.
Figure 1: Real GDP by sectors (in 2005 constant prices). Mining share in GDP is on the right hand side
Source: Appendix to Statistical Bulletin, December 2010 and 2011 Statistical Yearbook.
To the extent that the mining sector is becoming a dominant force in Mongolian economy
movements on the international markets are having greater influence (see the first panel of
Figure 2). Visually, prior to the collapse of the socialism in the late eighties and early nineties
the GDP growth was not that susceptible to the world copper price changes compared to more
recent experiences. As a tangible measure of its influence, the second panel of Figure 2 shows
the contemporaneous correlation between the real growth rate of the overall economy and the
copper price changes at the London Metals Exchange in blue horizontal dashes. On average
this correlation is at around 36%. The lagged effect shown in red horizontal line is even
stronger at around 40%. There is also some evidence that these correlations have not been
stable over the period of 1983-2011. Once we allow for possible changes in the correlations we
0
5
10
15
20
25
30
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
Agriculture Manufacturing Mining and quarrying
Net taxes on products Non-tradables Financial intermediation
Mining share (RHS, %)

ERI Discussion Paper Series No. 2
32
obtain the upward trending lines which suggest the contemporaneous correlation has almost
doubled since 19833.
Figure 2: Real GDP growth and copper price changes are plotted in the first panel. Contemporaneous
and lagged correlations between the real GDP growth and copper price changes are plotted in the
second panel. The horizontal lines assume the correlations to be constant while the upward-trending
lines allow for time variation.
Source: IMF Financial Statistics and NSO.
Such an upward shift can be explained in two ways. The first is an obvious one: as the
mining’s contribution to the GDP increases the more correlation is expected. It’s important to
note here that although the copper’s weight in the overall mining portfolio has been overtaken
by coal the correlation is still increasing, suggesting that everything in the portfolio is driven
by the same global factors.
But the mining share in GDP has never exceeded 30% and it’s around 20% in 2011. Rather
unobvious yet important another reason could be that the rest of the Mongolian economy is
positioning itself so that they are more dependent on the mining sector than ever before. This
would be extremely dangerous on at least two grounds. The first is the lack of diversification;
the mining sector pulling down the others in the event of weak global demand for raw
materials4. The second is from the arguments based on the Staple theory. Since the mining rent
is extremely high during the good times the cost structures of its suppliers and customers are of
little relevance to them. Then it loosens a basic market discipline for the other sectors in the
Mongolian economy since efficiency will not be encouraged and required of them.
Reflecting its boom, the mining’s share in export earnings rose from 50% in 1992 to 89% in
2011 after substantial and prolonged stagnation in late 90’s and early 2000’s due to the slump
in the demand for commodities (Figure 3). During this time the minerals earnings rose on
average 20% a year, with 50% increase just in 2011 reflecting the increased coal production;
even more growth is expected once Ouy Tolgoi, which is one of the largest undeveloped
copper and gold mines in the world, starts operating in 2013.
3 Here I use local projections by calculating correlations between the GDP growth and the copper price changes
on weighted data where the weights are centered at each year τ such that an observation at time τ + k has weight
λ_|k|, for k =...,-2, -1, 0, 1, 2, ... and λ = 0.9, with the results then plotted for each τ .
4 Two commercial banks ”Zoos” and ”Anod” turned into receiverships during the 2008-2009 global crisis as their
exposure to the mining was intolerably high.
-10
-5
0
5
10
15
20
-30
-10
10
30
50
70
90
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Copper price change (%, LHS) GDP growth (%, RHS)

Macroeconomic risks of Mongolia and ways to mitigate them
33
Figure 3: Real export income by products in 2011 million USD; Mineral products share in total export
is on right hand side
Source: NSO. US CPI is obtained from the Federal Reserve Bank of St Louise.
Cashmere textiles and textile articles contributed 21.4% of export income in 1992 but it
reduced to 5% in 2011, perhaps somewhat reflecting economic difficulties in its main markets:
European Union and Japan. Real export income from this sector rose 5.5% a year compared to
an average growth of 14.1% for the total export.
The composition of main mineral export earners is changing. Figure 4 illustrates export of coal,
which was almost non-existent prior to 2008, took over copper, which is traditionally the most
important export commodity. Gold, another traditional export commodity, is now in the sixth
place after iron, oil and zinc.
Companies associated with mining have played increasingly important roles in the state budget
income (Figure 5). Their tax contribution reached about 70% just before the global financial
crisis and it recovered to 56% in 20115. Real tax income from these companies rose 15 times
(37 times in nominal term) since 2002 which increased the total tax income 4.7 times (11 times
in nominal term). However tax income from the mining companies has been extremely volatile
as it declined 24% and 36% in 2008 and 2009 respectively in real terms. These large
fluctuations in revenue collections and associated complications in public debt management
will have an adverse impact on the economy. Without prudent fiscal management Mongolia
may be trapped between the boom and bust cycle: borrowing more than is warranted during the
good times and being unable to service the debt or paying too high interest rates during the bad
times.
5 Mongolian taxes are collected by three different authorities: Mongolian Tax Authority (MTA), responsible for
domestic taxes and levies, Mongolian Customs General Administration, in charge of taxes on foreign trade, and
General Authority for Social Insurance, that collects social security contributions. Tax income by economic
sectors is not reported on a regular basis and figures reported here are from MTA only.
0
15
30
45
60
75
90
0
1000
2000
3000
4000
5000
Mineral products Raw & processed hides, skins etc.
Textiles & textile articles Natural or cultured stones, precious metal, jewellery
Live animals, animals origin products Food, vegetable and base metal based products
Others Mining share (RHS, %)

ERI Discussion Paper Series No. 2
34
Figure 4: Export income from commodities ranked by their size (in thousand USD)
Economic growth, at least as fast as its population growth, is a necessary condition for the
sustainability of any society. However it is not a sufficient condition. Despite improvements in
growth, export and budget income the poverty, unemployment and income inequality issues
either still persists or are on the rise. Although the real growth rate between 1995 and 2010
averaged 5.3%, the poverty headcount actually rose from 36.3% to 39.2%. The number
dropped to 29.8% in 2012 perhaps thanks to the Human Development Fund’s cash transfer
program of 2010-2012 worth 19% of the 2010 GDP.
The mining boom is being generated by incredible price increases in most of the commodities.
However this could very well be just a short term phenomenon. Figure 6 plots the long term
copper price as a representative, both in nominal and real terms. Price of copper expressed in
terms of a basket of consumer products have reached the current heights only twice before the
beginning of the twentieth century but unlikely to continue for long as the high prices
encourage developments of both the new mines and cheaper alternatives. According to
Cuddington and Jerrett (2008) there is a super cycle in commodities prices and the current hike
is being driven by Chinese industrialization and urbanization, whereas earlier super cycles
were driven by similar developments in the United States, Europe, and Japan.
0.0
50000.0
100000.0
150000.0
200000.0
250000.0
300000.0
350000.0
400000.0
450000.0
500000.0
Coal Copper Iron ore
Oil Zinc ore and concentrate gold
processed cashmere molibdenium ore and concentrate Flour spar
Raw cashmere

Macroeconomic risks of Mongolia and ways to mitigate them
35
Figure 5: Real tax income in 2011 million MNT. Mining companies’ contribution to total tax income is
on the right hand side.
Source: Mongolian Tax Authority and Mongol Bank.
To put Mongolia’s current boom in a historical and international context Figure 7 illustrates the
dynamics of per capita Gross Domestic Product expressed in 2005 purchasing power parity
and real GDP growth by country groups along with that of Mongolia. Each country group is
based on 50 countries that best reflect that group.
Figure 6: Real (red dashed, RHS) and nominal (blue line, LHS) prices of copper. USD/ton
Source: U.S. Geological Survey, Federal Reserve Bank of St Louis and LondonMetal Exchange.
Mongolia is specifically compared to countries that depend on ore, minerals and fuel (which
includes coal, oil and gas) exports. Ore and mineral dependent countries suffered most from
stagnation in their living standards in the mid-seventies and even deterioration in the eighties
-
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
0
200000
400000
600000
800000
1000000
1200000
1400000
1600000
1800000
2000000
2200000
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Corporate income tax Royalty, land and water fees
Windfall tax on gold and copper and progressive royalty Domestic VA
T
Personal income tax Other domestic taxes
Other taxes and fees Non-tax income
Other income taxes Estate tax
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
Nominal Real

ERI Discussion Paper Series No. 2
36
and nineties. The gap between the living standards widened significantly during this period
compared to the richest and, more importantly, the most manufacturing dependent countries.
Fuel export dependent group suffered from a similar fate and clearly underperformed
compared to the world average.
The second panel of Figure 7 shows that from the mid-seventies to the midnineties the ore and
mineral dependent countries tended to grow slower than any other countries and suffered the
most from global recessions. The picture is reversed starting from the early nineties, perhaps
boosted later on by Chinese and India’s industrialization and urbanization.
Mongolia is currently enjoying a mining boom both in terms of new deposit discoveries and
high international prices mostly driven by the newly industrialized countries. The boom
created a revenue windfall, raising questions about an economy’s absorptive capacity and ways
to spend the excess revenue. Price volatility is likely to affect the fiscal balance of Mongolia
whose revenue mainly consists of commodity-related taxes, royalties, and dividend income.
When commodity prices drop, revenue will fall, forcing Mongolia to cut spending or incur
debt. To counter this risk Mongolia has recently established a stabilization fund, on paper, to
save commodity related revenue when prices are high and to draw money for budget support
when prices are low6.
Figure 7: Logarithms of per capita Gross Domestic Products in 2005 purchasing power parity are
plotted in the first panel. A country group is based on 50 countries that best reflect that group. Their 5
year weighted moving averages of real GDP growths are graphed in the second panel. The weights used
are 0.33, 0.77, 1, 0.77 and 0.33. Vertical shades represent the U.S. recessions dated by the NBER.
Source of information: World Development Indicators and NBER.
However, making such a fund a success story appears to be a challenge and the mere existence
does not guarantee a positive outcome. Fasano (2000) summarizes six economies’ experience
with resource funds and examines their contributions to the public financial management.
Davis, Ossowski, Daniel, and Barnett (2001) review the shortcomings of nonrenewable
resource funds and conclude that such funds are often part of the problem. However, given that
Mongolia’s economic growth, budget and export income largely depend on inherently volatile
mining sector it’s vital to have a larger buffer. This translates into enlarging the existing
6 Although given the resources are nonrenewable such a fund is often given a inter-temporal saving’s function.

Macroeconomic risks of Mongolia and ways to mitigate them
37
stabilization fund and clarifying its expenditure rules so that it cannot be used for purposes
other than its mandated role.

ERI Discussion Paper Series No. 2
38
3. Financial market risks
In this section I review some of the risks in the financial system. These include rapid credit
expansion in the presence of extremely high interest rates, decreasing intermediation rate and
debt overhang issues.
3.1 Excessive credit expansion
Mongolia has a quite liberal financial system and according to some is even ’overbanked’
(Marshall and Walters, 2010). Currently there are 14 privately owned and 1 state owned
commercial banks operating in Mongolia. The 14 banks accounted for 96% of total financial
system assets in September 2010 and unlikely to have changed much since. Three of them are
100% foreign owned and another one has foreign ownerships exceeding 50 percent (IMF
2011).
Figure 8. 2010 deposit rate (on the horizontal axis) distribution of the world economies is plotted in the
first panel. International deposit rates since the last Mongolian banking crisis of 1998 are graphed in the
second panel. Mongolian deposit rate is illustrated by a solid blue line and those of the other countries
are plotted by pink dots.
Source: IMF Financial Statistics.
Despite this Mongolia benefitted little from the financial liberalization in terms of the lending
rates. In fact Loukoianova (2011) finds high interest rates in Mongolia a puzzle. One reason of
high lending rate is the cost of attracting deposits. Mongolian deposit rate is quite high
compared to other countries with similar socio-economic conditions and those started the
transition to market economy at more or less the same time. A histogram of world’s deposit
rates is plotted in the first panel of Figure 8 where the horizontal line has annual deposit rates
before taxes and the vertical axis represents the frequency. With an annual deposit rate of
11.9% in 2010 Mongolia is in top 5 percentile of the world countries in terms of savings
nominal return, which was lower than just a handful of countries such as Yemen, Congo,
0
5
10
15
20
25
30
35
40
12345678910111213141516171819
Mongolia
Congo
Angola
Turkey
Venezuela
Yemen

Macroeconomic risks of Mongolia and ways to mitigate them
39
Venezuela, Turkey and Angola7. Deposit holders have also been benefitting from a tax
exemption on interest income that was introduced at the beginning of the transition period from
a centrally planned economy to a market oriented one with the purpose of encouraging savings
and supporting newly created banking sector. There is also a Government-funded blanket
deposit guarantee that started in 20088. Anecdotal evidence provided by the Mongol Bank
officials however, is that 5% of the holders, some of whom are non-residents, hold 95% of all
the term-deposits in Mongolia9. Such concentration of deposits can create a hostage-taking
situation for the individual banks as well as the whole banking sector, as the consequences of
possible capital flights are significant.
Relatively high deposit rate in Mongolia is not just a reflection of the fact that the other
countries have reduced their interest rates in wake of the word economic recession and
financial crisis. Comparatively high deposit rate is a persistent phenomenon in Mongolia as the
second panel of Figure 8 shows. Here the time series dynamics of the Mongolian deposit rate is
illustrated by a solid blue line and those of the other countries are plotted by pink dots. Rather
worrisome feature here is that the deposit rate have dropped significantly in early 2000 and had
been there for quite some time as if it has reached its ”natural rate”
Figure 9: Real deposit and lending rates
Source: IMF and Bank of Mongolia
7 The international comparison is for 2010, among 145 IMF member countries due to data avail-ability.
Mongolian domestic currency nominated, weighted average rate decreased to 10.5% in 2011 and increased to
11% in the second quarter of 2012.
8 The state guarantee coverage is reduced in June 2010. This includes an elimination of coverage for interbank
deposits, restrictions on deposits of related parties and limiting coverage on interest earned on deposits in excess
of the policy rate. However, as late as March 2011, MNT 3 trillion were covered by this blanket guarantee (IMF,
2011). Unless extended it will expire by the end of 2012.
9 An official recent statistic on savings holdings is that 52.2% of all deposits belong to 0.4% of the deposit holders
(Bayardavaa, Munkhbayar, Lkhagvajav, 2010).

ERI Discussion Paper Series No. 2
40
High deposit rate is likely to be the reflection of the inflation premium as inflation often makes
real rates turn negative (Figure 9). Unfortunately this is currently thecase and will likely to
reduce deposit and make funding more expensive in the near future.
Reflecting to some extent the high deposit rate loans have been quite expensive. With an
annual rate of 20.1% in 2010 Mongolia is ranked among the top 13 percentile of world
economies (the first panel, Figure 10)10. Again this is not just a manifest of the fact that other
countries have been implementing quantitative easing after the Global Financial Crisis, as can
be seen from the second panel of Figure 10.
At this high level of interest rates the loan decomposition is predictably geared towards
consumption, mining and quarrying and non-tradable sectors such as mortgages, wholesale and
retail trade, repair of vehicles, and construction where the returns are the highest during the
mining boom (Figure 11). These sectors accounted for three quarters of the outstanding loan by
the end of 2011. Manufacturing sector used to wield 32% of total loans but now accounts for
only a third of that which is mainly concentrated in capital intensive mineral processing plants.
As the mortgage comprises increasingly larger share of the loan portfolio the maturity of the
average loan is likely to increase. Proportion of real estate, construction and consumer loans in
the total loan portfolio increased from less than 5% in the first quarter of 2000 to about a half
in the second quarter of 201211. During which time loans with maturities greater than 5 years
have shot up from none to 75% for the real estate (for the total loan portfolio corresponding
increase was from 9% to 17%). Since most loan rates, especially those for the mortgages, are
fixed the stability in the banking sector depends on the stability in the deposit rates. Any future
hike in inflation will force the deposit rate to increase and exacerbate the maturity mismatch
problem. A recent estimate of the weighted average duration for the banking sector’s liabilities
is 3.8 months (Bayardavaa, Munkhbayar, Lkhagvajav, 2010).
Banking sector’s own response to the maturity mismatch problem was the establishment of the
Mongolian Mortgage Corporation (MMC) in 2006 that buys mortgages from the commercial
banks and then securitizes and sells them off to investors. The bad reputation its U.S. quasi-
government counterparts created after the Global financial crisis and the lack of strong
credibility in the absence of investment ratings for its securities will likely hinder its future
development as a solution to the mismatch problem. However its growth is not to be
underestimated. It had made 8 purchases of MNT 7.2 bln of mortgage loans from the
commercial banks by the end of 2011 (slightly less than 1% of the mortgage loans in the
banking sector). But as of Sep 2012 it claims to have 20 purchases worth MNT 30 bln which is
3.4% of the mortgage loans12. It’s not apparent how the mortgage loans are adjusted at the
Bank of Mongolia’s (BoM) bookkeeping after the deal between the MMC and commercial
banks. Moreover there is a potential conflict of interest, as the MMC is regulated by the
Financial Regulatory Commission yet the BoM is one of the MMC’s founders.
10 133 countries whose data were available are included in comparison. Mongolia’s weighted average lending rate
declined to 16.6% in 2011 but increased to 17.9% by June, 2012.
11 Mortgage loans used to be classified under consumer loans (the category Other) prior to Quarter 3 of 2008.
12 2011 MMC Annual Report and Anniversary Notice on Daily News 2012.9.4, p7.

Macroeconomic risks of Mongolia and ways to mitigate them
41
Figure 10. 2010 lending rate (on the horizontal axis) distribution of the world economies is plotted in
the first panel. International lending rates since the last Mongolian banking crisis of 1998 is plotted in
the second panel.
Source: IMF Financial Statistics.
Another response is that banks are increasingly trying to raise medium and long term funds
from the international market, the most recent one being the Trade and Development Bank’s 3
year, $300 mln bond issue with 8.9% coupon rate. Success of these initiatives however
depends on Togrog not depreciating during this period and/or availability of hedging.
Deposits are not only short term but also significantly denominated in foreign currency,
implying any devaluation in Togrog would entail exchange rate losses. Foreign currency
denominated accounts make up for almost a third of the current and savings accounts and
proved to be extremely volatile: within 5 months after July 2008 37% of the former and 16%
of the latter had been withdrawn. It’s also particularly alarming that the credit expansion is
happening in the exceptional period of low interest rates in the international markets. Ten-year
government bond yields are marked at historic lows at around 1.5% a year for the US, UK and
German bonds while that for the Japanese is below 1%. This temporary period of low interest
rates is sustained by two international factors that are crucial to Mongolia and clearly not
sustainable in the longer term. The first is the savings glut due to the baby-boomer retirement
savings and the reserves accumulated in China, as a result of its exchange rate interventions,
and in Arab countries following the run up in oil prices. But pensioners in Japan and Germany
have already started to draw down their savings. Prospects of hard landing in China and slower
demand growth for Arab countries’ export commodities might soon make them to cash out
their savings to support their economies. Secondly, the developed countries have been
implementing extensive quantitative easing to combat the global financial crisis and boost up
their economies. Future reversal on policy stance could increase interest rates sharply.
0
2
4
6
8
10
12
14
16
18
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
More
Congo
Madagascar
Kyrgiz
Yemen
Zambia
U
g
anda
Ta
j
ikstan
Sierra Lione
Para
g
ua
y
Malawi
Lao
Jamaica
Geor
g
ia

Figure 1
Others i
services,
administ
r
12% (in
selected
s
Risks a
s
interve
n
the ma
g
the big
g
for the
r
b
e achi
e
60% fo
r
involve
m
already
h
13 Intenti
o
such as
G
apartmen
t
the latter
r
14 30% m
o
1: The first
p
nclude loa
n
Finance, I
n
r
ation and
d
descending
s
ectors and
t
s
sociated
w
n
tions. Alth
o
g
nitude of
M
g
est. Howe
v
r
eal estate i
n
e
ved even
i
r
the rest
o
m
ent whic
h
h
igh in Mo
n
o
n of the ”10
0
G
egeenten and
t
s that could
b
r
esidential di
s
o
rtgage rate i
s
p
anel is the
n
s in Trans
p
n
formation
a
d
efense, Wat
order). Th
e
t
heir cumula
t
So
w
ith the m
o
o
ugh it’s n
o
M
ongolia’s
v
er good th
e
n
the prese
n
i
n the deve
o
f Europe
h
turned int
o
n
golia, wit
n
0
000 apartm
e
Khansvill ca
n
b
e bought wit
h
s
tricts, are ad
v
s
from Mong
o
ERI Disc
u
cumulative
t
p
ortation,
A
a
nd commu
n
er supply a
n
e
panels b)
t
ive shares i
n
urce: Mongo
l
o
rtgage ex
p
o
t unusual
f
”100000 a
p
e
intention
1
3
n
ce of sup
p
loped cou
n
and reach
e
o
sub-prim
e
n
essed by c
o
e
nt” program
i
n
be bought
w
h
the Govern
m
v
ertising thei
r
o
l bank’s loan
u
ssion Paper
S
42
t
otal loans
o
A
griculture,
n
ication, E
d
n
d Professio
n
and c) rep
o
n
the total lo
a
l
Bank Quarl
e
p
ansion ar
e
f
or the go
v
p
artment p
r
3
is, the 6
%
p
ly
b
ottlen
e
n
tries; hom
e
e
d 69% in
e
mortgage
o
mmercial
b
i
s not clear a
s
w
ith the progr
a
m
ent loan so
m
p
roperties as
report 2012
Q
S
eries No. 2
o
utstanding i
n
Housing a
n
d
uca
t
ion, H
e
n
al services
o
rt proportio
n
a
n portfolio.
e
rly Loan Rep
o
e
being ele
v
v
ernments t
o
r
ogram” w
o
%
mortgage
p
e
cks. Unive
r
e
ownershi
p
the US a
f
crisis. Giv
e
b
ank mort
g
s
even luxuri
o
a
m loan. Alt
h
m
e
b
uilding c
”can be merg
Q
2.
n
real term
(
n
d food ser
v
e
alth, Electr
i
which acco
u
n
s of mid t
o
o
rt.
v
ated thro
u
o
support t
h
o
uld certain
p
rogram is
r
sal home
o
p
is 40% f
o
f
ter the ex
t
e
n the dem
a
g
age rates a
s
o
us apartmen
t
h
ough there is
ompanies, in
c
ed afterwards
(
in descendi
n
v
ices, Adm
i
i
city and g
a
u
nt for the
r
o
long ter
m
u
gh the go
v
h
e first tim
e
ly qualify
a
increasing
o
wnership
c
o
r German
y
t
ensive go
v
a
nd for mo
r
s
high as 3
0
t
s in resident
i
a size restric
t
c
luding the o
n
”.
n
g order).
i
nistrative
a
s, Public
r
emaining
m
loans of
v
ernment
e
buyers,
a
s one of
demands
c
ould not
y
, around
v
ernment
r
tgages is
0
%
14
, any
i
al districts
t
ion for the
n
e building

additio
n
financia
l
is also
c
supply
1
5
school,
b
uyers
a
increase
even pe
r
Figure 1
2
first pan
e
since the
Housin
g
infrastr
u
transpo
r
networ
k
high on
commo
n
costlier
would f
o
In contr
a
rate ha
s
services
without
cost of
f
15 Within
2800 app
l
apartmen
t
(Daily N
e
n
al subsidy
l
situation
f
c
ertain to c
r
5
. Althoug
h
primary, s
e
a
nd constru
c
d provisio
n
r
haps in the
2
. 2010 inte
r
e
l. It's the d
i
last Mongo
l
g
problem
s
u
ctural net
w
r
t link con
n
k
s), making
the agend
a
n
practice i
n
to buy an
d
o
rce some
a
a
st to a rel
a
s
been con
t
that woul
d
a doubt th
a
f
unding. A
the last 5 mo
l
ications and
t
t
s, to the gov
e
e
ws, Aug 21,
2
Macroeco
n
on the de
m
f
or not onl
y
r
eate enviro
h
giving hi
g
e
condary,
v
c
tion secto
r
n
of the chil
constructi
o
r
mediation r
a
i
fference be
t
l
ian banking
s
hould be
m
w
orks, imp
r
n
ecting th
e
the reside
n
a
. Moreove
r
n
the worl
d
d
hold a se
c
a
bsentee ho
m
a
tively quic
k
t
inuing. A
l
d
not jeop
a
a
t the curre
cursory gl
a
nths since th
e
t
he Develop
m
e
rnmen
t
-own
e
2
012).
n
omic risks o
f
m
and side
w
y
those who
nments of
c
g
her priori
t
v
ocational
a
r
compared
dcare coul
d
o
n sector.
a
te (on the
h
t
ween the le
n
crisis of 19
9
Source: I
M
m
ore effec
t
r
ovement
o
e
city sub
u
n
tial buildi
n
r
careful in
c
d
, could als
o
c
ond or a t
h
m
eowners
t
k
fall and t
h
l
though th
e
a
rdize the f
i
nt competi
t
a
nce at the
e
Housing Fi
n
m
ent Bank all
o
e
d State Ban
k
f
Mongolia a
n
43
w
ill just inc
r
are not eli
g
c
orruption
a
t
y to real
e
a
nd tertiar
y
to the lon
g
d
also incre
a
h
orizontal a
x
n
ding and d
e
9
8 are plotte
d
M
F Financia
l
t
ively dealt
o
f public tr
u
rbs to th
e
n
g land all
o
c
lusion of
h
o
ease the
p
h
ird home
a
t
o sell their
h
en flatteni
n
e
level of
”
i
nancial sy
s
t
ion can an
d
intermedia
t
n
ancing Corp
o
o
cated MNT
2
k
, only comm
e
n
d ways to mi
r
ease the p
r
g
ible but al
s
a
s the dem
a
e
state deve
l
y
levels) g
i
g
-term soci
a
a
se the lab
o
x
is) distribut
i
e
posit ra
t
es.
d
in the seco
l
Statistics.
with from
ansports (
e
e
centre v
i
cation syst
e
h
omes in th
e
p
rice hikes.
a
s a specul
a
vacant flat
s
n
g of the de
p
”
natural”
o
s
tem stabil
i
d
perhaps
s
t
ion rate th
a
o
ration starte
d
2
0 bln, tha
t
i
s
e
rcial bank t
h
tigate them
r
ice level f
o
s
o who end
a
nd for suc
h
l
opments
o
i
ves imme
d
a
l returns o
f
o
r participa
t
i
on of the w
Internation
a
nd panel.
the suppl
y
e
.g. introdu
c
i
a upgradi
n
e
m more tr
e
existing
r
A recurrin
g
a
tive bet o
n
s
or rent the
p
osit rate t
h
o
r healthy
i
ty is yet t
o
s
hould only
a
t is define
d
accepting a
p
s
sufficient to
h
at was willi
n
o
r all, wors
e
up receivi
n
h
mortgage
s
o
ve
r
educat
i
d
iate gains
f
the huma
n
t
ion in the
s
orld econo
m
a
l intermedi
a
y
side. Ext
e
c
tion of fa
n
g the exi
s
ansparent
s
r
eal estate t
a
g
tax woul
d
n
rising pri
c
m to the p
u
h
e decline i
n
margin on
o
be deter
m
go so far
g
d to be a
d
p
plications th
e
buy approxi
m
n
g to distribut
e
e
ning the
n
g one. It
s
outstrip
i
on (pre-
to home
n
capital,
s
hort run,
m
ies in the
a
tion rates
e
nsion of
st public
s
ting rail
s
hould be
a
x law, a
d
make it
c
es. That
u
blic.
n
lending
banking
m
ined it’s
g
iven the
d
ifference
e
y received
m
ately 400
e
the loans

ERI Discussion Paper Series No. 2
44
between the lending and the deposit rates reveal that the current margin is not unusually large
compared to other countries. Cross sectional histogram as shown in the first panel of Figure 12
Mongolia is in the 31st percentile and the second panel shows that there is still a downward
trend.
Therefore relatively higher rates in Mongolia could be a reflection of her inherent risk
premium for inflation and exchange rate fluctuations, and not because her banks charge too
much for their services. But the explosion in credits in the presence of high interest rates does
point to i) increased exposure of the whole banking sector to the inherently volatile mining and
non-tradable sectors where the rent is sufficiently high, ii) possible deterioration in the quality
of the loans.
The share of nonperforming loans, a popular banking system indicator, has substantially
declined overall from the peak of 23% in the third quarter of 2009 to 6% in the second quarter
of 2012 after taking two commercial banks into receiverships in the process16. These two
bankruptcies are thought to be caused by their excessive exposures to mining industry, with the
proportion of non-performing loans increasing unprecedentedly in the last three quarters of
2008. In fact non-performing loans almost tripled for the consumer and wholesale-retail loans
and doubled for the real estate within three quarters. Mongol bank’s two main policy
instruments have been tightened recently but the impact is yet to be seen. The reserve
requirement was increased twice and policy rate was raised five times since the first quarter of
2010 yet the total credit grew by 78% in real term since then.
Therefore, the amount and speed of concentration of the consumer and real estate related loans
should be capped through prudential banking regulations. It’s recommended to require
commercial banks to put aside additional reserves on the basis of maturity mismatch between
its deposits and loans. Unified customer loan database and deposit insurance system should be
implemented as soon as possible. It’s also vital to keep inflation at low and stable level to
enable financial deepening.
3.2 Debt overhang
Mongolian debt dynamics from 1991 to the first quarter of 2012 is shown in Figure 1317. Prior
to 2006 only public debt information is available, as the Mongol Bank started to publicize
detailed information only in 201018.
The level of investment loans, which was almost non-existent before 2008, increased 1.4 times
in 2009, tripled in 2010, again increased 5.4 times in 2011 and now constitutes 60% of all debt.
These loans are mainly concentrated on big mining projects and are likely to increase even
further. Mongol Bank reiterates that mining companies could internalize these investment
loans.
16 Here non-performing loan includes past due in arrears, substandard, doubtful and bad loans.
17 Public debt information is from IMF (2005) and IMF (2008) for the periods 1991-2001 and 2002-2006
respectively. The rest is from the Mongol Bank.
18 It’s assumed that other sectors had no access to international debt markets.

Macroeconomic risks of Mongolia and ways to mitigate them
45
Figure 13: Mongolian external debt dynamics in thousand USD
Source: Mongol Bank and IMF.
Government sector’s borrowing rose 31% in the first quarter of 2012 after the Development
Bank’s bond sales. The debt is on commercial terms which are more stringent than the long
term concessional loans that Mongolia used to be eligible for as a low-income country. The
management of this new type of debt will complicate macroeconomic management and
increase country’s risk level, especially when the government’s repayment ability is affected
by the commodity price volatility.
The confluence of degrading quality of institutions and the availability of the external funding
due to the short term potential income gives the authority a dangerous leverage, if used
inappropriately that could harm the long term growth of the economy. There is an ample
historical experience of debt overhang problems in Latin American countries.
Some legal guards against Mongolian governments accumulating too much debt is within the
Fiscal Stability Law to be fully implemented in 2013. These include a cap on public debt (not
to exceed 40 percent of GDP in net present value), a structural budget deficit ceiling of 2% of
GDP, and speed limit on budget expenditure growth (not to exceed non-mining GDP
growth)19. There are some questions that need to be answered before implementing the law
such as what discount rate to use in the NPV calculation, why the expenditure is restricted by
non-mining GDP growth (given that non-tradable sectors are likely to grow faster than the
mineral GDP). Moreover Mongolian government is engaged in multiple activities that may
create uncertain demands on future fiscal resources and, therefore, complicate fiscal policy.
These include loan guarantees, loans taken for the purpose of on-lending to state owned
enterprises and local governments, Public Private Partnerships, and the Development Bank of
Mongolia. These activities and related risks are claimed to be growing in terms of type,
complexity and size.
Some debt instruments could address this concern by linking debt payment to commodity
prices and the GDP or its growth rate. Coupons and principal payment of commodity-linked
bonds are linked to a stated amount of a reference commodity. Because the volume is fixed, a
country’s debt payment is positively related to its export commodity prices; as a result, its debt
burden declines following the plunge of commodity prices. O’Hara (1984) studied the use of
19 For more views on Mongolian Fiscal Stability Law see Isakova, Plekhanov, and Zettelmeyer (2012).

ERI Discussion Paper Series No. 2
46
commodity bonds to stabilize consumption. Claessens (1991) pointed out that commodity
bonds can be used to hedge debt management problems associated with volatile export
earnings. However, the pool of investors willing to have exposure to commodity risks is
smaller than those invest in traditional bonds.
Stable and profitable functioning of the Development Bank of Mongolia should be maintained
as its potential to affect Mongolia’s financial sector standing on the international market is
significant. Therefore, payments associated with its bonds could be linked with GDP growth or
commodities to reduce its risk.
To conclude this section the main risk of debt overhang is not in the private sector, as some
recent statistics might indicate, but in the government sector. I see the foreign investors’
reluctance not to internalize their investments as equity but keeping them as a loan is some
form of a ”defense mechanism” against the rent seeking as the investment project failure will
impact on Mongolia’s international credit rating. A stable legal environment where these
investors can operate for longer periods will hopefully induce them to change strategy.

Macroeconomic risks of Mongolia and ways to mitigate them
47
4. Portfolio risk of Mongolian export basket
The markets for minerals, oil and its related products, and many other commodities are
characterized by high levels of volatility. This is mainly associated with the production
sluggishness, therefore almost vertical supply curve, in the short run. This volatility is claimed
to have a negative impact on economic growth, export volatility, income distribution, and
poverty alleviation (Larson, Varangis, Yabuku, 1998). Dawe (1996) calculates instability
indices for a cross-section of countries by taking account of the share of exports in any given
economy and finds that export instability is negatively associated with growth and investment.
Hausman and Gavin (1996) found that volatility decreased economic growth and investment
and adversely affects income distribution and raises poverty rates in Latin America.
Precisely because commodity markets are volatile, hence risky, governments, producers and
consumers seek ways of managing and transferring risk. In response to this need, markets for
commodity risk trading arose, and their use has become increasingly widespread. Instruments
traded in these markets include futures and forward contracts, options, swaps, and other
derivatives (Kletzer, Newbery and Wright, 1990). Rolfo (1980) investigated the use of futures
for cocoa producing prices and provided a framework to calculate the optimal hedge ratio in
the presence of both production (output) and price volatility. He showed that a limited usage of
the futures market for hedging would be optimal when there is production variability.
However, because of the huge losses incurred on derivative markets in mid nineties by traders
at the Bank of Mongolia, the subject nowadays is almost a taboo. This is however in contrast to
the international trend of increased government participation in derivative markets despite
important barriers to access including counterparty risk, and basis risk (no correlation of local
and international prices).
Figure 14: Commodity price dynamics in USD, January 1979-May 2012
Source: IMF Financial Statistics and http://www.indexmundi.com.
For a country that depends on a basket of commodities the volatility can be moderated if there
is so called a portfolio effect, i.e. if price fluctuations of commodities offset each other.
However Pindyck and Rotemberg (1990) find that many commodity markets experience short
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
0
20
40
60
80
100
120
140
160
180
200
1979m01
1980m02
1981m03
1982m04
1983m05
1984m06
1985m07
1986m08
1987m09
1988m10
1989m11
1990m12
1992m01
1993m02
1994m03
1995m04
1996m05
1997m6
1998m7
1999m8
2000m9
2001m10
2002m11
2003m12
2005m1
2006m2
2007m3
2008m4
2009m5
2010m6
2011m7
Coal (metric ton, lhs) Oil (barrel, lhs)

ERI Discussion Paper Series No. 2
48
sharp price spikes followed by extended periods of considerably lower prices at the same time.
Mongolia is not only becoming heavily dependent on mining industry as described in the
Background section but also its export basket is concentrating on fewer commodities.
Therefore it’s important to assess the extent to which Mongolian export is diversified and its
evolution over time.
Figure 15: Time varying residual correlation coefficients of VAR(1) model estimated over February
1973 to March 2012. The results are obtained by estimating VAR models with weights centered at each
month τ such that an observation at time τ + k has weight λ|k|, for k=...,-2, -1, 0, 1, 2, ... and λ = 0.99,
with the results then plotted for each τ .
As can be seen from the prices of the four main commodities plotted in Figure 14 they have
different statistical characteristics such as growth rates and variability. It would be ideal for
Mongolia if the prices of its commodities grow over time, with as little fluctuations as possible.
Moreover since commodities market is known for its fluctuations it’s ideal if the price drops of
some commodities are offset by increases in others. To gauge the extent of such counter-
movements in commodity prices I run a simple VAR and calculate the correlation coefficients
of its residuals in Figure 15. When a local projection analysis is carried out there appears
hardly any counter movements but an increase in contemporaneous correlation among them in
recent years.
Price dynamics of Mongolian export commodities are plotted in four panels of Figure 16 along
with their international counterparts. Also illustrated in the graphs are prices of futures with
maturities that go as far into the future as they are currently sold in New York Metal Exchange.
These futures contracts, along with forward contracts, are among the most important
instruments for risk management and their markets should provide a convenient way for
Mongolia to reduce risk20. The longest maturity of futures contract was December 2019 for
20 A forward contract is an agreement to deliver a specified quantity of a commodity at a specified future date, at a
price (the forward price) to be paid at the time of delivery. The commodity specifications and point of delivery (as
well as the quantity, price, and date of delivery) are spelled out in the contract. There are two parties to a forward

Macroeconomic risks of Mongolia and ways to mitigate them
49
copper, June 2017 for gold, July 2016 for copper, and December 2014 for coal in December
2011.
Using Markowitz’s (1952) mean/variance portfolio allocation analysis on four main export
commodities, which are copper, gold, coal and oil, the rest of this section assesses the risk
dynamics of Mongolian export income21. As of May 2012 these four commodities made up
74% of total export income (NSO Statistical Bulletin).
If the production of these commodities is not restricted the optimal composition of the export
basket can be determined22. Although this condition is unlikely to be fully satisfied in reality
we describe the main gist of the Markowitz’s methodology as it will be used for a modified
purpose. Let xt be a vector of portfolio weights, normalized to sum to unity, of the export
income and E(rt) be the vector of expected increase in export income from the commodities,
i.e., ],...,[ ,,1 ′
=tntt xxxand ])(),...,([)( ,,1 ′
=tntt rErEE r. Then the expected increase of the
portfolio value is
,∑,,′

 .
The variance of the portfolio is
,∑,

 2∑∑ ,



 ,,,′.
Here Σ is a variance covariance matrix with variances of the changes in commodity incomes
on the diagonal and covariances on the off-diagonal positions. So the optimization problem is
to find that xt which maximizes a risk adjusted increase of the portfolio value; that is ,
,
where c is an arbitrary number. By varying c one could find all the possible optimal portfolios
(which is referred to as efficient frontier in finance). The maximization should satisfy certain
conditions, such as xt not to contain negative numbers, which all add to unity, to say the least.
contract: the buyer (or long position), who will receive the commodity and pay the forward price, and the seller
(or short position), who will deliver the commodity. Forward contracts are often traded directly among producers
and industrial consumers of the commodity; in some cases they are traded on organized exchanges (such as the
London Metals Exchange). A futures contract is also an agreement to deliver a specified quantity of a commodity
at a specified future date, at a price (the futures price) to be paid at the time of delivery (dotted lines in Figure 16).
Futures contracts are usually traded on organized exchanges, such as the New York Mercantile Exchange or
London Metal Exchange, and as a result, tend to be more liquid than forward contracts and their data are more
readily available. Other than this, a futures contract differs from a forward contract only in that the futures
contract is ”marked to market,” which means that there is a settlement and corresponding transfer of funds at the
end of each trading day.
21 Mineral products constituted 89.2% of total export in 2011. Other major mineral products that are not included
in this analysis due to the lack of sufficiently long time series data on their prices include iron ore, zinc,
molybdenum and flour spar.
22 As the reviewer rightfully pointed out that the export portfolio’s composition should not be restricted by one
single sector but should be derived as part of the entire wealth of the nation, including non-traded assets such as
human capital. This indeed could be very interesting line of future research. But from a practical perspective,
passive managers even in advanced economies with better data availability segment the market into various asset
classes, such as stocks and bonds, or into even finer classes such as large and small stock (Sharpe, Alexnader,
Bailey, 1999, pp.232). This paper is obviously not the first one to use the optimal portfolio methodology in a
single sector. For example, Love (1979) and Bertinelli, Heinen, Strobl (2009) evaluates the benefits to export
diversification using this approach while Larkin, Sylvia, Tuininga (2003) uses it even for optimal seafood
processing.

ERI Discussion Paper Series No. 2
50
It’s informative to plot the realizations of expected increase in income, , against its
associated risk measure , .
Figure 16: Spot prices of Mongolian gold (kg), oil (barrel), coal (ton) and copper (ton) in USD,
along with their international counterparts. Futures prices are plotted by pink dots.
A. Gold price B. Oil price
C. Coal price D. Copper price
Source: IMF financial Statistics, DataStream and NSO.
Rather than optimizing with respect to t
x here I use the actual export income composition of
Mongolia for t
x and track the portfolio’s value increase with its associated risk over time. For

Macroeconomic risks of Mongolia and ways to mitigate them
51
that we need estimates of the means, variances and covariances. Equation (1) provides with a
VAR model that is used to estimate these statistical measures.
t
p
i
itit urAδr++= 
=
−
1
(1)
where Ai (i = 1, …, p) are n×n coefficient matrices and δ is a n×1 vector of intercepts. The
error term t
u in (1) has mean zero and variance-covariance matrix E(utut') = Σ, and is
temporally uncorrelated. Further, defining D to be the diagonal matrix containing the standard
deviations of t
u and P to be the corresponding correlation matrix, then by definition Σ = D P
D. The methodology proposed in Bataa et al. (2013a) dates structural breaks in each of the
three components of equation (1): that is, firstly, the VAR coefficients Ai (i = 1, …, p) and δ,
which together capture mean effects as well as dynamics; secondly, (conditional) volatility
measured by D; and finally, contemporaneous (conditional) correlations in P. In addition to
dating any breaks found to occur in equation (1), I also examine the statistical significance of
commodity relations by conducting inference on Ai and P. To conserve space the reader is
referred to the above paper for the details of the testing methodology.
Unconditional mean of the growth rates is given by
δAIμ
-1
1







−= 
=
p
i
i (2)
When the system of equations in (1) is estimated on data from January 1979 to March 2012
and tested for structural breaks I find no breaks in the mean coefficients but 3 breaks in
covariance matrix. WDmax test statistic for the null hypothesis of no breaks against unknown
number of breaks up to 5 in the conditional mean assumes 39.89, which is lower than the 5%
critical value of 47.08. However the same statistic is statistically significant for the covariance
matrix as the test statistics 198.5 is much higher than the corresponding critical value of 28.72.
Since the null of no breaks is rejected it’s important to identify the number of breaks and their
locations. The sequential test statistics that identify the number of breaks rejects the null of one
break against two breaks (the statistic 206.83 is higher than the critical value of 30.29) and the
null of two breaks is rejected against three breaks (the statistic is 44.95 and the critical value is
31.44). It was impossible to continue this further because of the lack of the data that satisfy the
methodology’s requirements. Thus I conclude there are 3 breaks.
The identified structural breaks in the volatilities and comovements of the commodities prices
are dated at September 1981 and June 1991 and April 2000 in Table 1. It also provides the 95
confidence intervals for these dates.
Table 2 provides the Granger causality test results. First of all there is statistically significant
persistence (or inertia) in the price movements of all commodities ranging from 0.26 for oil
and coal to 0.17 for gold. More interestingly it’s found that copper Granger causes oil and oil
Granger causes coal. The latter link, although weak in magnitude, is understandable as they are
often used as substitutes for heating. A 10 percentage point increase in oil price leads to 0.6
percentage point increase in coal prices. But the 10 percentage point increase in copper prices

ERI Discussion Paper Series No. 2
52
indicates 1.3 percentage point increase in oil prices in the following month. This portfolio
effect is an interesting finding given that the mining uses a large amount of oil products which
are currently imported from Russia23.
Table 1: Iterative Structural Break Test Results. It shows asymptotic WDmax and asymptotic sequential
test statistics resulting from Bataa et al.’s (2013a) procedure, with the latter comparing l + 1 versus l
breaks, beginning with l = 1, with the respective 5 percent critical value in brackets. N.C. indicates that
the test cannot be computed because no additional break can be inserted while satisfying the minimum
regime length requirement. Estimated break dates (in bold) are based on the iterative bootstrap
algorithm; this is followed by the 90 percent confidence interval for this date. The value in parentheses
is the bootstrap percentage p−value for the specific break at convergence. The VAR lag lengths is p = 1
chosen by HQ information criteria.
Coefficients Covariance Matrix
Asymptotic WDmax Test Statistics [and Critical Values]
39.89 [47.08] 198.50 [28.72]
Asymptotic Sequential Test Statistics [and Critical Values]
206.83 [30.29]
44.95 [31.44]
N.C.
Break Dates, Confidence Intervals (and Bootstrap p-values)
1981.09
1980.11
1981.11
(0.00)
1991.06
1991.03
1992.03
(0.00)
2000.04
2000.01
2001.05
(0.00)
Conditional on the break dates in Table 1 and Granger causality test results in Table 2 the
origins of the covariance matrix breaks are examined in Table 3. At all three break dates the
volatility changes. The correlation structure also changes at the first two break dates but the
break date on April 2000 is not statistically significant at 5 percent significance level. But the
finding is only marginal. Looking at the correlation numbers in the current regime one can see
that the synchronous movements have never been so strong in the past (especially so if one
looks at the correlation matrices defined by all the covariance matrix breaks in Table 4). The
null hypothesis that the price movements of oil and coal are not correlated with any of the
other commodities used to be accepted with high probability of 73% and 46% in the seventies
and early eighties, but not anymore. Although these high and positive correlations imply
amplified income generation during the price increase it also means a drying-up of income
during bad times.
23 See Bataa (2012b) for detailed and current information on Mongolia’s fuel import and consumption situation
with respect to mining sector.

Macroeconomic risks of Mongolia and ways to mitigate them
53
Table 2: Individual Coefficient Estimation Results. Columns represent equations. The first value in
each cell reports the estimated coefficients of the VAR. The values in parentheses are bootstrap p-
values (expressed as percentages) for the null hypothesis that the corresponding true value is zero. **
indicates significance at the 5% level and * significance at the 10% level, both using the bootstrap p-
value.
E
xplan.
V
ariables
Dependent Variable
Copper Gold Coal Oil
I
ntercept 0.24** 0.40** 0.17** 0.24**
(0.0) (0.0) (0.0) (0.0)
Copper 0.34** -0.03 0.06 0.13**
(0.0) (54.3) (12.7) (4.5)
Gold 0.05 0.17** 0.03 -0.03
(43.5) (1.0) (55.4) (69.7)
Coal 0.04 0.01 0.26** 0.04
(57.0) (87.5) (0.0) (56.5)
Oil 0.01 0.02 0.06** 0.26**
(74.7) (44.7) (2.8) (0.03)
Tables 4 and 5 provide all the necessary information for the analysis of the portfolio analysis,
which is summarized in Figure 17. Vertical axis represents the historical average price
increases and the horizontal axis has volatility, which is a proxy for risk, measured in standard
deviations. Here red dots indicate the risk-return profile of the individual commodities.
Obviously gold is the best commodity as its value has the least historical fluctuation and has
been steadily increasing since 1979. In contrast copper and coal, the most important
commodities for Mongolian economy, are characterized by the most price fluctuations; the
price trend has been better for copper. Black plus sign indicates the portfolio’s risk and
expected increase for a given year. I track the optimality of export portfolio since 2000 due to
the break in covariance matrix in 2000. Portfolio has the same characteristics as copper that
year since the portfolio is almost fully consists of it. The following year gold increases its share
to a third as a result of the ”Gold Program” implemented by the government of Mongolia. It
has to be said that a substantial amount of tax and non-tax favors were generously handed out
to attract investors. The portfolio risk and return improved significantly. That situation is
further improved until 2006 when gold’s contribution is receded to 28%, 20 percentage points
decrease from a year earlier. This deterioration in portfolio performance is improved in 2008
before falling back again. Since then coal has been dominating the portfolio and will be so in
the near future given the developments of OT and TT; it has some unfavorable historical
performance as can be seen from the graph and Mongolian economy’s risk level will be
heightened.

ERI Discussion Paper Series No. 2
54
Table 3: Volatility and Correlation Results. Notes: Panel A reports the significance of
structural break tests for the diagonal elements of the covariance matrix of the VAR
(Volatility) and for the off-diagonal elements of the correlation matrix (Correl), showing
bootstrap p-values (expressed as percentages) for the test of no change over adjacent
Covariance Matrix subsamples identified in Table 2, with the result placed against the dates of
the later subsample. The values reported are the final ones computed in the respective general
to specific procedures of Bataa et al. (2013a). The corresponding sub-sample residual standard
deviations are reported in Panel B and subsample contemporaneous residual correlations in
Panel C. The standard deviations and correlations are computed after merging subsamples
based on the respective break test results in Panel A (using 5% significance). The final column
of Panel C reports the bootstrap p-value for a test of the joint hypothesis test that all
contemporaneous correlations relating that commodity are zero. All results are obtained from a
VAR in which the restrictions implied by the results of coefficient breaks and
persistence/dynamic interaction tests (at 5% significance) are imposed.
A. Si
g
nificance
of Breaks
B. Subsample Residual
Standard Deviations C. Subsample Contemporaneous Correlations
Subsample Volatility Correl. Copper Gold Coal Oil Copper Gold Coal Zero
Correlatio
n
72.02 –81.09 5.43 7.23 4.01
8.
29 Copper 0.1
Gold 44.15 0.1
Coal -5.49 -6.15 73.2
Oil -4.86 10.08 7.62 45.6
81.10 –91.06 0.0 0.7 6.27 2.54 2.61
7.
43 Copper 3.5
Gold 13.33 0.9
91.07 –00.04 0.0 0.0 4.21 3.34 4.57
8.
34 Coal 24.7 5.28 1.5
Oil 3.41 29.99
-
15.24 0.3
00.05 –12.03 0.0 5.6 8.87 4.84 8.82
8.
23 Copper 0.0
Gold 29.95 0.1
Coal 24.39 16.97 0.0
Oil 39.97 14.28 26.91 0.0
Blue curve indicates optimal portfolios when there is flexibility of production; that is the best
level of return for a given level of risk had Mongolia the choice over the weights of its export
basket. Experiences of the last 13 years indicate however Mongolia is diverging further from
such a concept.
It’s true that once a project starts the Government has no right to dictate the company about
how much to produce. However the government has the right to choose what type of licenses
to give away to the mining companies. When making such decisions the government could use
similar analysis as here. Moreover this framework could be extended to become more dynamic
such that the means, the variances and the covariances are predicted beforehand and used for
forecasting. Alternatively, one could refine the data into monthly frequency and create a
framework of precautionary buffer accumulation dependent on the estimated level of risk.

Macroeconomic risks of Mongolia and ways to mitigate them
55
Table 4: Quantities for export portfolio. Notes: Panel A reports the expected price growth obtained by
equation (2) and imposing the Granger causality test results (at 5% significance). Panel B reports the
estimated variance-covariance matrices where the diagonal elements are the variances and the off-
diagonal elements are the covariances, over the sub-samples. The corresponding sub-sample
contemporaneous residual correlations are reported in Panel C.
Subsample
A.
Expected
growth
B. Variance-Covariance matrix C. Correlation matrix
Copper Gold Coal Oil Copper Gold Coal
72.02 –81.09 Copper 0.38 29.26
Gold 0.48 17.14 51.84 44.15
Coal 0.28 -1.08 -1.81 16.08 -5.49 -6.15
Oil 0.40 -2.05 5.94 2.66 68.2 -4.86 10.08 7.62
81.10 –91.06 Copper 0.38 38.99
Gold 0.48 2.13 6.78 13.33
Coal 0.28 4.01 0.34 6.78 24.70 5.28
Oil 0.40 1.59 5.76 -2.94 54.8 3.41 29.99 -15.2
91.07 –00.04 Copper 0.38 17.6
Gold 0.48 3.84 11.03 27.60
Coal 0.28 3.38 2.78 20.64 17.72 18.42
Oil 0.40 8.26 1.59 6.7 69.1 23.52 5.80 17.7
00.05 –12.03 Copper 0.38 77.57
Gold 0.48 14.55 23.88 33.96
Coal 0.28 26.56 6.5 76.88 34.34 14.50
Oil 0.40 46.31 10.95 29.15 66.8 64.30 27.11 40.5

ERI Discussion Paper Series No. 2
56
Table 5: Export income from four commodities. Notes: Panel A reports the income generated by the
export of four commodities. Panel B expresses individual commodities contribution towards a portfolio
consisting of these four. Panel C reports the share of this portfolio in Mongolian total export income.
A
. Commodities income in thousand USD
B
. Commodities share in the portfolio C. Portfolio
in total
export
Copper Gold Coal Oil Copper Gold Coal Oil
2000 160,275.8 0.0 0.0 1,812.9 0.99 - - 0.01 0.35
2001 147,998.1 74,748.8 0.0 1,777.7 0.66 0.33 - 0.01 0.43
2002 140,232.1 117,644.7 0.0 3,092.8 0.54 0.45 - 0.01 0.50
2003 163,694.6 137,648.3 0.0 4,567.7 0.54 0.45 - 0.01 0.50
2004 284,322.0 242,242.3 0.0 6,279.7 0.53 0.45 - 0.01 0.61
2005 326,216.7 331,410.6 26,625.6 9,261.6 0.47 0.48 0.04 0.01 0.65
2006 635,420.2 270,105.1 45,640.6 19,841.1 0.65 0.28 0.05 0.02 0.63
2007 811,502.5 234,873.6 116,226.4 53,403.1 0.67 0.19 0.10 0.04 0.62
2008 835,666.0 599,883.0 184,665.9 101,936.6 0.49 0.35 0.11 0.06 0.68
2009 501,923.7 308,473.2 306,300.6 115,632.5 0.41 0.25 0.25 0.09 0.65
2010 770,594.0 178,339.0 881,998.3 154,386.1 0.39 0.09 0.44 0.08 0.68
2011 963,596.0 113,046.6 2,250,046.
4
252,191.8 0.27 0.03 0.63 0.07 0.75
2012, I-V 339328.1 38,859.1 803,181.1 127,157.8 0.26 0.03 0.61 0.10 0.77
Figure 17: Optimality of the Mongolian export portfolio composition. Red dots indicate the historical
average price increases and their associated volatility, measured in standard deviations (which are
proxies for risk). Black plus sign indicates the portfolio’s risk and expected increase for a given year.
Blue curve indicates optimal portfolios when there is flexibility of production.
Export income increase
Risk (σ)
0.28
0.33
0.38
0.43
0.48
4.4 5.4 6.4 7.4 8.4
2000
copper
coal
oil
gold
2005 2004
2012 2011
2010
2009
2008
2007
2003
2006
2001
2002

Macroeconomic risks of Mongolia and ways to mitigate them
57
5. Risks induced by government macroeconomic policies
5.1 Pro-cyclicality of Mongolian macroeconomic policies
Economists agree that macroeconomic policies should be directed to smooth out the business
cycle fluctuations; that is counter-cyclical24. It’s important to determine the actual stance of
fiscal and monetary policy because what matters to the economy is the confluence of their
impacts. Therefore wrong identification of the business cycle stance might result in a stronger
or weaker response than is warranted by the Ministry of Finance and the Bank of Mongolia.
It’s often argued that fiscal policy is pro-cyclical in Mongolia yet monetary policy is regarded
as countercyclical. The criticism is however not only specific to Mongolia but also to all
developing countries (Gavin and Perrotti, 1997). Kaminsky, Reinhart and Vegh (2004) report
some evidence that Mongolia has the most pro-cyclical fiscal policy in terms of inflation tax
among 104 countries they studied.
Various theoretical models have been proposed to explain this phenomenon and many of
which seem to be relevant for Mongolia. The most widely shared explanation is the credit
market imperfection. Developing countries can borrow at relatively affordable interest rates
only when their economies are doing good allowing them to increase the public spending and
run deficits. When their economic prospects are bleak the credit market is shut so they have to
cut spending when they are supposed to do the opposite.
Another strand of theory is based on political economy models. Alesina, Tabellini and
Campante (2008) argue that when the economy is booming and the government is assumed to
be corrupt the public demands more public goods, lower taxes or cash transfers. This is
because the public do not observe the rents generated by the boom and fear the isappropriation
by the corrupt elites. As a result the government increases borrowing more than is warranted
and run deficits when the economy is already overheating. Hence the procyclicality is the
result of an agency problem.
To determine the extent to which macroeconomic policies are procyclical and whether that has
been changing over time I use the following models. The first model is Model 2 of Ilzetzki and
Vegh (2008)25:
ttt yg
ε
β
+= (3)
tttt ugyy ++= −
φ
α
1 (4)
where t
g and t
y are cyclical components of government spending and output, t
ε
and t
u are
i.i.d. white noise with 0)( =
tt uE
ε
. The cyclical components are obtained from the quarterly
24 The opposite type of policy to the counter-cycle one that ”adds fuel to fire” is termed pro-cyclical and a policy
in between that stays neutral to the business cycle fluctuations is called acyclical.
25 More comprehensive analysis is the subject of an ongoing project but early results show that conclusions
reached here remain the same.

ERI Discussion Paper Series No. 2
58
data using the Hodrick Prescott filter with a smoothing parameter of 1600. The data coverage
spans from Q1 of 1998 to Q2 of 201226. If 0>
β
fiscal policy is procyclical, if 0=
β
, fiscal
policy is acyclical, and if 0<
β
, fiscal policy is countercyclical. Since the government
expenditure is a part of the GDP, equation (4) will be used to assess the size and trend of its
contribution. If
φ
> 0 then the government expenditure contributes to the cyclical fluctuations
and detracts otherwise. To account for more possible lagged effects of output on government
spending I use heteroskedasticity and autocorrelation consistent standard error estimates.
In terms of monetary policy’s stance over the business cycle and its evolution I use the
following version of Taylor rule from Kaminsky et al. (2004); which they borrowed from
Clairida, Gali and Gertler (1999):
ttt yi
λ
π
π
θ
γ
+−+= )( (5)
where t
i is the policy controlled short term interest rate (here I use the weighted average of
Central Bank Bill rate),
π
π
−
t represents deviations of actual inflation from its sample
average. If 0>
λ
, monetay policy is countercyclical, if 0=
λ
, monetary policy is acyclical,
and if 0<
λ
, monetary policy is procyclical. Again I use heteroskedasticiy and autocorrelation
consistent standard errors for inference.
All specifications require a measure of output gap while equations (3) and (4) also use a
deviation of the expenditure from its long term trend. Here I use Hodrick and Prescott (HP)
filter for consistency, with a smoothing parameter of 1600. Prior to applying the HP filter I
deseasonalize all the time series as in Bataa (2013a), that implements an iterative procedure
suggested in Bataa et al. (2013b). Going back to estimation of equation (1), ideally one has to
use policy instrument variables rather than outcome variables to assess fiscal policy stance, as
argued in Kaminsky et al. (2004). Since the central government’s consumption data that
excludes non-discretionary components such as loan interest repayments and transfers were
available only from 2000 I also complement this with the current expenditure, capital
expenditure and their combinations.
To assess whether macroeconomic policy stance changed over time I use a simple technique of
local projection as a descriptive analysis. The results are obtained by estimating models for (3)
and (4) with weights centered at each month τ such that an observation at time τ+k has weight
λ|k|, for k =...,-2, -1, 0, 1, 2, ... and λ = 0.99, with the following results then plotted for each τ .
To be specific, for each τ a general form of the model, either (3) to (4), is estimated and plotted
against τ (Figure 18).
Final results corresponding to the fiscal policy is reported in the first panel of Table (6) which
shows strong evidence of pro-cyclicality. When the output increases from its long term trend
by 1 percentage point the expenditures also increase by 1.29 to 3.61 percentage points,
depending on the type of expenditures, from their trends in the following quarter. The strongest
cyclicality is observed for the capital expenditures and the weakest cyclicality is present in the
26 Since government spending decisions can plausibly be made with only a certain lag I assume here that the most
recent economic condition affects the current expenditure; thus two equations are estimable by OLS as the error
terms in both equations are uncorrelated with the right hand side variables. More comprehensive analysis of
cyclicality is the subject of more focused future research.

current
paymen
t
Figure 1
policy’s
(red hori
are plott
e
Local p
r
cyclical
i
There is
fluctuat
i
govern
m
Moneta
r
regressi
o
27 The re
v
”Consum
p
This cou
l
combine
d
expenditur
e
t
s and trans
8: Pro-cycl
i
contributio
n
zontal lines
)
e
d by blue d
o
r
ojection g
r
i
ty might b
e
also a sig
n
i
ons in the
G
m
ent expen
d
r
y policy e
q
o
n non sig
n
v
iewer wond
e
p
tion+ capita
l
l
d be due th
e
d
. Moreover, t
h
Macroeco
n
e
, perhaps
fers. All re
g
i
cality coeff
i
n
to the cycl
i
)
. Local pro
j
o
tted curves.
r
aphs in t
h
e
on the ris
e
n
ificant rev
e
G
DP growt
h
d
itures on g
r
q
uation rev
e
n
ificance c
a
e
red why R2
l
expenditure
”
e
seasonal p
r
h
e governme
n
n
omic risks o
f
due to th
g
ressions a
r
i
cient for fi
s
i
cal fluctuati
j
ection coun
t
h
e first col
u
e
.
e
rse causali
t
h
. Interesti
n
r
owth.
e
als insigni
a
nnot be r
e
increased su
b
”
compared t
o
r
operties of
t
n
t consumptio
f
Mongolia a
n
59
e inclusio
n
r
e highly si
g
s
cal policy
on of the o
u
t
erparts of t
h
u
mn of Fi
g
t
y: expendi
t
n
gly, local
p
ficant coef
f
e
jected. Ac
b
stantially in
o
the cases o
f
t
he individua
l
n is 8 quarter
n
d ways to mi
n
of non-c
y
g
nificant
27
.
is plotted i
n
u
tput growt
h
h
ese coeffic
i
g
ure 18 mo
t
ure cyclic
a
p
rojections
a
f
icient of 0
.
cording to
the cases o
f
f
”Current ex
p
l
series are
b
shorter than
t
tigate them
y
clical ite
m
n
the first
c
h
is graphed
i
ents that al
l
reover sug
g
a
lity contri
b
a
lso reveal
i
.
176 and th
this result
f
”Current+ c
p
enditure” an
d
b
eing differe
n
t
he rest due to
m
s such a
s
c
olumn and
in the seco
n
l
ow for tim
e
g
est that t
h
b
uting to th
e
i
ncreasing
i
h
e null hyp
o
monetary
apital expen
d
d
”Capital ex
p
n
t from whe
n
data availabi
s
interest
the fiscal
n
d column
variation
h
ese pro-
e
cyclical
i
mpact of
o
thesis of
p
olicy is
d
iture” and
p
enditure”.
n
they are
l
ity.

ERI Discussion Paper Series No. 2
60
acyclical. Failure to find significant relationship between the output fluctuation and the
monetary policy response could be the use of the Central Bank Bill rate as a proxy for the
policy variable. Mongol bank started to announce policy rates only in 2010 the use of which
resulted in highly insignificant estimation.
Table 6: Estimation results for equations 3-4. Columns 2-4 report estimated coefficients. ** indicates
significant at 1% . Also reported are the regression R2 and F statistic for the regression significance.
β
φ
λ R
2 F
A. Procyclical fiscal policy
Current expenditure 1.29**
7.83**
Capital expenditure 3.61**
9.07**
Current+capital expenditure 2.02**
16.26**
Consumption+ capital
expenditure 2.17**
15.58**
B. Output growth
Current expenditure .081** .170 5.54**
Capital expenditure .034** .174 5.68**
Current+capital expenditure .104** .248 8.90**
Consumption+ capital
expenditure .103** .275 8.91**
C. Countercyclical monetary policy
C. Central Bank bill rate .176 .07 2.28
From the estimations one can see that the government expenditure growth is playing
increasingly larger role in the overall economic fluctuations. Therefore it should be
recommended to take into account the exiting business cycle measurements such as output gap
before implementing these government programs. Moreover, the government initiatives in
fighting corruption, enforcing contracts and improving institutional quality are expected to
have broad impact.
5.2 Inflation and Real Appreciation of Togrog
Money supply has been showing a boom and bust cycle. M2 grew by 56% in 2007, before
shrinking by 5.5% in the following year, and increased again by 26.9% and 62.5% in 2009 and
2010 before moderating to 37% in 2011. The 2010 increase was the fastest in the world and
similar growth was observed only in 1994 (79%) as a part of the “shock therapy” transition
from the centrally planned economy to a market oriented one.
Figure 19 illustrates a well documented long run relationship between money growth and
inflation using all the available data since 1961. Mongolia seems to follow this relationship
well having 23% inflation in 2008 and already 10.6% by July 2012. Prospect of this higher
inflation do not necessarily translate into real exchange rate appreciation if the nominal
exchange rate is kept flexible.

Macroeconomic risks of Mongolia and ways to mitigate them
61
But the Central Bank appears to prefer maintaining the nominal exchange rate stability,
although the effectiveness of such policy has been controversial. The exchange rate is one of
the automatic stabilizers for commodity dependent countries, appreciating during the boom,
thus reducing export and depreciating during the bust thus encouraging export, but Mongolia
has been quite adamant using this buffer. Figure 20 plots the dynamics of the so-called
commodity currencies, i.e. those of countries that heavily depend on commodities (Cashin,
Cespedes and Sahay, 2004).
During the early phases of the Great Financial Crisis of 2008, global economic perspectives
became bleak and commodity prices declined. Central banks of commodity currency countries
accepted depreciation given the negative terms of trade shocks originating from the global
factors. In contrast, Mongolia defended its currency by selling from its official foreign reserve
to maintain the exchange rate. Given that the length of such policy cannot be maintained for
extended period of time this invites some form of speculative move: borrow in Mongolian
Togrog, buy U.S. dollars from the Central bank and hold them until the eventual currency
depreciation, when the initial loan is repaid.
Figure 19: Relationship between the average money growth and average inflation. There are 178
countries; beginning dates differ. Logarithms were applied to ease comparison.
Source: World Bank.
After losing a half of its foreign reserve of 1 billion USD, Mongolia devalued its currency by a
quarter and approached IMF with a request of Stand-By arrangement with amount of USD 450
million and obtained the loan. This was of exceptionally large size compared to its quota28.
28 A related policy risk appears to be the lack of ownership of macroeconomic policies in Mongolia. Mongolia had
implemented several IMF programs before the 2008 request, since joining the organization in 1991 and
immediately receiving a Stand-By program. In 1993 Mongolia made a three year arrangement under the Enhanced
Structural Adjustment Facility (ESAF) in an amount equivalent to SDR 40.81 million. Another three year
-1
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Money growth (1961-2010)
Inflation (1961-2010)
Mongolia (2010)
Congo, D.R.
Brazil
Angolia
Peru
Bolivia
Ukrain
Argentin
Azerbaijan
Belarus
Kazakhstan
Armenia
Mongolia

ERI Discussion Paper Series No. 2
62
Figure 20: Commodity currency dynamics by countries. USD rate on 2007.6.29 is set to unity. If the
rate moves above it that means exchange rate appreciation and vice versa.
Source: Bloomberg.
Figure 21 shows the evolution of official foreign exchange reserves of commodity dependent
countries. The June 2007 amount is set to unity. As can be seen from the graph Mongolia has
the most activist policy. After the devaluation of late 2008 the commodity prices recovered so
did the values of the commodity currencies. Mongolia preferred to build its foreign exchange
reserve aggressively that was almost depleted earlier, rather than allowing its currency to
appreciate to fend off the commodity boom-bust cycle. Sterilization process that followed may
have contributed to inflation, sustained domestic interest rates at high levels, and stimulated
even greater capital inflows and worsening the credit affordability. However it’s worth
noticing that the exchange rate has become relatively flexible after the crisis, as that was one of
the conditions of the Stand-By program. By the third quarter of 2011 the official reserve had
almost tripled compared to its pre-crisis level. Interestingly, when the Euro crisis deepened in
the third quarter of 2011 and commodity prices declined as a result of that, Mongolia repeated
its previous mistake by not allowing its currency to adjust to this new reality. Togrog did
depreciate but far more sluggishly than others and that meant when the others had already been
stabilized Mongolian currency was still continuing its slide29.
arrangement under ESAF/PRGF in an amount equivalent to SDR 33.4 million was approved in 1997. Again in
2001 a three-year arrangement under the PRGF in an amount equivalent to SDR 28.49 million was approved in
2001. But IMF involvement has been shown to effect the economic growth negatively (e.g. Przeworski and
Vreeland 2000, Hutchison 2003, and Dreher 2006). The IMF can influence economic outcomes by its money, the
policy conditions attached to its loans and, more generally, its policy advice. Barro and Lee (2005) argued that
greater involvement in IMF programs tends to lower the rule of law and democracy and conclude that a typical
country would be better off economically if it committed itself not to be involved with IMF loan programs. Dreher
and Walter (2010) find however that the IMF does help the crisis stricken countries to preserve exchange rate
stability and correct macroeconomic imbalances.
29 Given the Mongol Bank’s independence is relatively weak this discretionary policy environment
has a danger of attracting possible political influences, hence increases investment uncertainty.
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
29/06/2007
29/08/2007
29/10/2007
29/12/2007
29/02/2008
29/04/2008
29/06/2008
29/08/2008
29/10/2008
29/12/2008
01/03/2009
01/05/2009
01/07/2009
01/09/2009
01/11/2009
01/01/2010
01/03/2010
01/05/2010
01/07/2010
01/09/2010
01/11/2010
01/01/2011
01/03/2011
01/05/2011
01/07/2011
01/09/2011
01/11/2011
01/01/2012
Canada Australia Chile Russia Mongolia South Africa

Macroeconomic risks of Mongolia and ways to mitigate them
63
Figure 21: Official reserve dynamics of commodity-currency countries. June 2007 amount is set to
unity. Movements above one mean reserve accumulation compared to the base date and vice versa.
Source: Bloomberg.
Keeping nominal exchange rate stable does not translate into real exchange rate stability. Real
exchange appreciates if the nominal rate appreciates and also if the inflation is higher than in
other countries. As Figure 22 shows Togrog’s nominal value in terms of US dollar and a basket
of currencies Mongolia trades with (NEER) has depreciated. But the real effective exchange
rate has been appreciating with only significant reversal in early 2009 due to the nominal
depreciation and is now still 50% stronger compared to 2000, hurting non-mineral exports and
tradable sectors in Mongolia.
Figure 22: Exchange rates. December 2000 is set to 100. Upward movements mean appreciation.
Source: Mongol Bank.
Mongol Bank was under pressure to defend the currency, and strangely, a protest rally was organized
by the Mongolian Trade Union, counterparts of which in other countries organize the same rallies
when exactly the opposite happens.

ERI Discussion Paper Series No. 2
64
6. Mining dependency, macroeconomic risks and institutions
Institutions are vital for understanding macroeconomic risks as they determine what the
policies would be. Good institutions create an environment that promotes economic activity,
inventiveness, growth, and development. Bad institutions typically result in economic
stagnation. Mongolia’s vast mineral and fuel resources promise major economic benefits, but
experiences in Africa and Latin America suggest that an abundance of such appropriable
resources can lead to poor policies and hinder long-term development; this is the so-called
”Resource curse” phenomenon.
In an American Political Scientists’ Society Meeting its president is said to have claimed that
Mongolia is an outlier that the existing political system theories could not explain (Rossabi,
2005). Fish (2001) described Mongolia as a remarkable puzzle among the post-communist
countries in terms of its political experience. While most of the countries in the Central Asia,
which had similar social-economic backgrounds, slipped into personalistic rule and
authoritarianism” Mongolia was the only post-communist country outside of East Europe to
receive a rating that entitled it to a classification as a free polity”. He attributed Mongolia’s
success in democratization to five main features: lack of natural endowments, low strategic
value for external players, a modesty not to become a great or regional power, an absence of
national father figure and a dilution of central power.
These all changed now. Mongolia has now discovered huge mineral and fuel deposits and has
already started relying on extractive industries for economic growth. According to Fish’s
(2001) prediction this would entail enormous political dangers and Mongolia may quickly
become little more than a battleground for actors seeking control over the resource rents. The
following two sub-sections provide some further discussions.
6.1 Institutional quality and macroeconomic risks
Resource rich countries have faced political problems ranging from greater levels of corruption
and rent seeking to internal conflict (Collier and Hoeffler, 1998, 2004, Lane and Tornell, 1999
and Torvik, 2002). The issue is of significant importance for Mongolia: governance problems
are more visible than in the past and corruption and rent-seeking seem to be on the rise.
Institutional quality is found to be the key to avoiding the resource curse (Mehlum, Moene, and
Torvik, 2006). Knack and Keefer (1995) conclude that the rule of law is a better measure of
institutional quality than others such as Gastil indices (measures of democracy) and political
violence measures. Adherence to the rule of law is manifested by maintenance of property
rights and absence of corruption. Maintenance of property rights is considered one of the most
important pillars of a free market economy and an important determinant of economic growth
through their effects on the level of investment (North, 1990; North and Weingast, 1989).
There is also overwhelming evidence that indicates corruption is detrimental for economic
growth (See e.g. Shleifer and Vishny, 1993; Ehrlich and Lui, 1999)30.
30 See Knack and Keefer, 1995; Mauro, 1995; Sala-i-Martin, 1997; for more evidence supporting the idea that
some measure of the rule of law or property rights or corruption is significantly correlated with growth of per

Macroeconomic risks of Mongolia and ways to mitigate them
65
Figure 23: Heritage foundation and World Bank institutional quality indicators. They range between 0
and 100 and the higher the better.
Source: 2012 Index of Economic Freedom and Worldwide Governance Indicators.
Unfortunately, exactly these measures of Mongolia’s institutional quality have been
deteriorating in the recent years, as can be seen from the World Bank and Heritage Foundation
indices illustrated in Figure 2331. In particular, the World Bank’s control of corruption index
has shot down within just a decade: the country ranked at 57th percentile in 2002 but the latest
index is at mere 27th (the higher the rank the better). Heritage Foundation’s index also points
to similar amount of deterioration. The World Bank’s rule of law measure, however picked up
slightly in 2011, shows a resounding downward trend since the mining boom started.
Figure 24 illustrates the dynamics in the Heritage Foundation’s freedom from corruption and
property rights indices of the world countries. Countries above the red line are those whose
indicators improved and those below the line are the losers. In 1996 Mongolia shared more or
less the same position with world average countries such as Spain, Ireland, Chile and Cyprus
for the corruption control and Switzerland, France and Sweden for property rights, perhaps
thanks to its relatively liberal laws and regulations. But the control of corruption has almost
collapsed in Mongolia, as witnessed by the fastest drop in the index in the sample of 181
capita real income, based on various institutional quality measures such as those in International Country Risk
Guide (ICRG) data and/or World Bank supported Kaufmann, Kraay, and Mastruzzi (2010) project data.
31 World Bank’s control of corruption measure reflects perceptions of the extent to which public power is
exercised for private gain, including both petty and grand forms of corruption, as well as ”capture” of the state by
elites and private interests while its rule of law measure reflects perceptions of the extent to which agents have
confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property
rights, the police, and the courts, as well as the likelihood of crime and violence. Both are percentile ranks among
all countries (ranges from 0 (lowest) to 100 (highest) ranks). Heritage foundation’s protection of property rights
measures the degree to which a country’s laws protect private property rights and the degree to which its
government enforces those laws. It also assesses the likelihood that private property will be expropriated and
analyzes the independence of the judiciary, the existence of corruption within the judiciary, and the ability of
individuals and businesses to enforce contracts. Its freedom from corruption index relies upon Transparency
International’s Corruption Perception Index that measures the extent to which corruption prevails in a country.

ERI Discussion Paper Series No. 2
66
countries only after two countries, Italy and Argentina, since the Heritage Foundation started
publicizing the index. In terms of property rights Mongolia’s deterioration is only exceeded by
Argentina, Saudi Arabia, Thailand and Venezuela among the world’s 181 countries. Failing to
reverse this trend in institutional deterioration would result in not only the failure to tackle all
the macroeconomic risks discussed in this paper, but also endanger social stability in
Mongolia. Another disturbing sign is the recent change of election rule; from being a
majoritarian into mixed one. The new rule, which allows 28 out of 76 Members of the
Parliament getting elected through the Party list and was used for the 2012 Election, opened
doors for those with good connections with party leaders and/or contributed most financially.
But Anderson and Aslaken (2008) show natural resources are more likely to reduce growth
when exactly this type of proportional electoral systems are in place than when the electoral
systems are majoritarian. Moreover Kunikova and Rose-Ackerman (2005) show the
proportional or mixed election rules tend to create environments where corruption and rent-
seeking flourish compared to the majoritarian system
Figure 24: Dynamics in Heritage Foundation indices. Each dot represents a country
Source: World Bank.
Thus, although the advantages of the new proportional election rule adopted in Mongolia could
be significant compared to the earlier rule, the country needs to rethink in terms of the current
economic reality where decisions about her point resource based rents being made by ever
decreasing number of political elites.
6.2 Resource dependency and prospects for higher unemployment
Currently important labor market adjustments are happening in Mongolia. Agriculture has been
dominating Mongolian labor market since the collapse of the socialism in early 1990’s and still
is the biggest single employer. Figure 25 shows the dynamics of the Mongolian labor force by
economic sectors.
0
20
40
60
80
100
020 40 60 80 100
1996
2012
Freedom from corruption
Ire la nd
Qatar
Chile
Urug ua y
Estonia
Cyprus
Spain Bhutan
Costa Rica
Belize
Papua New Guinea
Kenya
Nigeria
ArmeniaMongolia
0
20
40
60
80
100
020 40 60 80 100
1996
2012
Property rights
Belize
Mongolia
SwitzerlandSweden
France
Estonia
Spain
St.Vincent
Portugal
St.Lucia
Isra e l Czech
HungaryBotswana
Namibia
Gambia
Uganda
Pakistan

Since t
h
mining
a
2011 (a
n
commu
n
employ
m
Public
a
uninterr
u
to its p
r
4% of
t
intensit
y
Figure 2
5
order ac
c
differen
c
without
c
In contr
a
b
een st
r
respecti
v
2001, r
e
transitio
This ex
p
the reso
u
intensiv
e
expecte
d
potentia
l
11.6%,
9
32 Only e
x
33 Bersch
and Qua
h
real GDP
.
years wit
h
from a V
A
As can b
e
h
e NSO sta
r
a
nd quarry
i
n
nual gro
w
n
ication se
c
m
ent was t
h
a
dministrat
i
u
pted grow
t
r
e 2009 lev
e
t
he total a
n
y
.
5
: First pan
e
c
ording to
t
c
e between
t
c
reating infl
a
a
st, the agr
i
r
uggling t
o
v
ely in the
l
e
versing th
e
n related u
n
p
ansion of
m
u
rce pull e
f
e
agricultu
r
d
to rise. T
h
l
output, i
s
9
.9% and
7
x
ception for t
h
and Sinclair
(
h
’s (BQ, 1989
.
Here I repli
c
h
only three e
p
A
R on real G
D
e
seen from Fi
g
Macroeco
n
r
ted produ
c
i
ng separat
e
w
th of 1.9
%
c
tor due to t
h
h
e second
i
on and d
e
t
h
32
while
e
e
l. Notwit
h
n
d is not e
x
e
l shows th
e
t
heir size i
n
t
he actual a
n
a
tionary pres
i
culture an
d
o
retain it
s
l
ast 17 yea
r
e
earlier m
i
n
employme
m
ining an
d
f
fects of the
r
e and ind
u
h
e current
m
s
largely p
o
7
.7% in th
e
h
e unabated g
r
(
2011) comp
a
) method as
i
c
ate their resu
l
p
isodes of th
e
D
P growth a
n
g
ure 25 BQ’s
n
omic risks o
f
c
ing a com
p
e
ly from th
e
%
). The hig
h
e birth of
I
fastest and
e
fense shr
u
e
mploymen
t
h
standing t
h
x
pected to
e
number o
f
n
2011. Th
e
n
d potential
sures and p
o
d
manufact
u
s
employ
m
r
s. Employ
m
i
gration tre
n
nt.
d
non-trada
b
Dutch dis
e
u
stry erode
s
m
easure of
t
o
sitive sin
c
e
last three
r
owth in the
p
a
re various m
e
i
t yielded mo
r
l
t with a long
e
e
economy
p
e
r
n
d inflation
w
output gap is
f
Mongolia a
n
67
p
rehensive
l
e
manufact
u
hest incre
a
I
T an
d
mo
b
other non
-
u
nk almost
t
in the mi
n
h
is growth
t
rise signif
i
f
employees
e
second pa
n
output. A
z
o
sitive gap i
n
u
ring, whic
h
m
ent: the e
m
ent in the
n
d from th
e
b
les at the
e
e
ase pheno
m
s
the natur
a
t
he output
g
c
e late 20
0
years (Fig
u
p
ublic sector
e
e
asures of out
p
r
e robust res
u
e
r sample. B
Q
r
forming und
e
w
ith a lag len
g
reasonably i
d
n
d ways to mi
l
abor class
i
u
ring the l
a
se was in
t
b
ile phone
c
-
tradable e
m
by 9% i
n
n
ing grew b
y
t
he mining
i
cantly in
t
in thousand
s
n
el shows
a
z
ero gap me
a
n
dicates the
e
h
make up
mploymen
t
agricultur
a
e
urban to
e
xpense of
t
m
enon. As
t
a
l rate of u
n
g
ap, the di
f
0
9 but the
u
u
re 25, sec
o
e
mployment p
r
p
ut gaps for
M
u
lts whether i
t
Q
output gap
h
e
r its potenti
a
g
th of 4, selec
t
d
entifying dz
u
tigate them
i
fication in
a
bo
r
force i
n
t
he transp
o
c
ompanies.
T
m
ployment
n
2011 re
v
y
28% in 2
0
employme
n
t
he long ru
n
s
. They are
a
n estimate
a
ns that the
e
conomy is
o
Mongolia’
s
t
grew on
l
a
l sector sta
r
rural areas
t
he tradabl
e
t
he competi
n
employm
e
f
ference be
t
u
nemploy
m
o
nd panel)
3
r
ior to that w
a
M
ongolia and
t
’s applied to
h
as been larg
e
a
l. This outpu
t
t
ed with Aka
i
u
d’s of 1999
a
1994 that
t
n
creased b
y
o
rtation, sto
T
he rise in
is also on
v
e
r
sing its
0
11 almost
r
n
t accounts
n
due to i
t
ranked in d
e
of the outp
economy i
s
o
verheating.
s
tradable s
l
y 2.1% a
n
r
ted to dec
l
to escape
e
s is consis
tiveness in
e
nt in the c
o
t
ween the a
c
m
ent level
h
3
3
. Accordi
n
a
s in 1998.
recommends
mineral or n
o
e
ly positive in
t
gap measur
e
i
ke Informati
o
a
nd 2001 as w
t
reats the
y
37% in
rage and
mining’s
the rise.
trend of
r
eturning
for only
t
s capital
e
scending
ut gap, a
s
growing
e
ctor has
n
d 2.5%
l
ine since
from the
tent with
the labor
o
untry is
c
tual and
h
as been
n
g to the
Blanchard
o
n-mineral
the last 15
e
is derived
o
n Criteria.
ell as 2009

ERI Discussion Paper Series No. 2
68
well known Okun’s law any mechanic attempt to reduce unemployment when the economy is
above its potential results in nothing but overheating. This reflects itself in price increases
including that for labor. Although the merit of the law is often debated the basic idea is hard to
dismiss: improve the quality of and access to the education.
Figure 26: The first panel plots the number of families involved in herding by herd size (LHS) and the
number of livestock per family (RHS). The second panel graphs herder dynamics (in thousands, LHS)
by age decomposition and the number of herders in a herder family (RHS).
Source: NSO Statistical Yearbooks.
This labor market adjustment is however increasing the vulnerability in the agricultural sector.
The first panel of Figure 26 illustrates the number of herder families in terms of their herd-size.
It reached a peak of 301474 in 1992 after the collapse of socialism and related closures of
factories in urban areas. Since then the number of herding families declined significantly,
perhaps with faster speed during the dzuds, to 211743 in 2011. Families with unsustainably
few or, economically unprofitable livestock tended to leave herding before 2007 but that trend
seems to have stalled perhaps due to the lack of opportunities elsewhere. But the youth defies
this trend and is progressively leaving herding as witnessed by their reduced proportion of 41%
compared to 58% in 1995 (second panel of Figure 26).
However this decline is happening against the rising number of livestock, thus increasing the
vulnerability of this sector in the absence of corresponding technological and/or organizational
improvements. The number of livestock was 25.7 mln in 1992 and reached 44 mln by 2009,
which translated into more than 190 livestock per herding family and more than 16 animals per
population. Such a large number of livestock was never heard of in Mongolia even during the
centrally planned economy when it kept at 12 livestock per person (Boone, 1994). This
increase conflated with a shift in herd composition in favor of pasture-hostile goats, due to
their valuable cashmere, brought with them over-grazing problems and sustainability issues (so
called tragedy of the commons). Large number of herds neither meant more productive
livestock sector nor cheaper meat as slaughtered meat weight per livestock has been on the
recession as below potential periods. Interestingly it also picks up a near potential output in the first quarter of
2011. I use quarterly data from the first quarter of 1998 to the second quarter of 2012 that have been seasonally
adjusted according to Bataa (2013a). When the VAR model is subjected to structural break test as in Bataa et al.
(2013a) no break is detected.

Macroeconomic risks of Mongolia and ways to mitigate them
69
decline since 1991 and the real mutton price has been on the rise (Figure 27). The record large
number of livestock per person in 2008 coincided with a record high meat price in real terms.
It appears that whenever the number of livestock exceeds the long term average of 12 per
person it corrects itself (Figure 27). The herd size build-ups of 1994-1998 and 2005-2009
reversed by the loss of approximately 11 million adult animals in each case during the dzuds
(extremely cold winters). The build-up seems to be happening again, witnessed by the soaring
meat prices and sharp increase in livestock per family in 2011. Universal cash transfers,
subsidization of the tuition fees and monthly allowances for the students, including those from
rural areas, are contributing to this trend as they act to deter economic engagements of
otherwise self sufficient herders. Current disconnect between the herders and the state should
be restored through taxes in return for private ownership of pasture land, improved access to
markets and better public goods.
Figure 27: Selected agricultural indicators
Source: NSO Statistical Yearbooks.
Lack of engagements in market transactions translates into the lowest growth of income
opportunities for the agricultural sector. Figure 28 shows the real wage dynamics in Mongolian
economy by sectors. Real wages have been increasing significantly throughout the economy,
spearheaded by the financial sector where the wages have increased almost 14 times since the
last quarter of 2001. Mining wages have been growing fast and are predicted to takeover those
in the financial sector. This will increase the wage pressure for the rest of the economy as the
employment size is almost three times bigger than the latter. Such a large sudden increase
reflects a skills mismatch among the graduates Mongolian educational system is providing and
the mining sector’s demand for qualified staff on a timely basis.
This may entail a further loss of competitiveness in the tradable sectors such as manufacturing
and agriculture. Real wages in Chinese manufacturing has been increasing over the years yet

ERI Discussion Paper Series No. 2
70
their monthly wage is very competitive compared to what’s offered in Mongolia34. In fact
when their hourly wage is converted into Mongolian Togrog using 160 working hours per
week (Mongolian normal employment hours) only agriculture, hotels and trade sector has
cheaper labor compared to the Chinese manufacturing sector.
Figure 28: Real wages in thousand MNT per month (expressed in 2012.Q2 ? using Mongolian CPI).
These are ordered according to their latest level. Chinese manufacturing hourly rate is multiplied by
160 hours a month and converted to MNT using the Yan/MNT exchange rate and weighed using
Mongolian CPI. It’s also deflated using Mongolian CPI to allow comparison.
However Mongolia needs competitive manufacturing, that is claimed to have many social
fringe benefits (Matsuyama, 1992) and is able to accommodate large amounts of labor. If the
current trend is left on its own, the natural rate of unemployment will go up as the mining is
not labor intensive. Given that Chinese manufacturing wage is extremely competitive it’s hard
to see Mongolia enter into labor-intensive manufacturing with low-skills requirements.
However improved quality and accessibility of education at all levels compared to what’s
currently offered in Mongolia and elsewhere in the region can increase the chances of high-
skilled manufacturing and/or knowledge based economy. Increased flexibility of foreign
workers entering Mongolia’s labor market and seasonal nature of the mining and construction
sectors are likely to pose further challenges in maintaining the economic competitiveness of
the Mongolian workforce without jeopardizing the current pace of expansion.
34 Banister, J. and Cook, G. China's employment and compensation costs in manufacturing through 2008.
Monthly Labor Review, March 2011, 39-52.

Macroeconomic risks of Mongolia and ways to mitigate them
71
6.3 Income inequality, weak institutions, and resource dependent economy
Mongolian population had relatively equal opportunities just after the collapse of the socialism,
like any other post communist countries that started the transition to the market system more or
less simultaneously. As the economy shifts from a labor intensive agricultural sector not to its
preferred next stage of manufacturing based economy but to capital intensive one more
unemployment and income divergence is expected. Moreover a point-based nature of the
resources will contribute even further to this trend.
But this widening gap in income and opportunities among the population might endanger the
social and economic stability. Mongolia’s latest measured Gini coefficient of income
inequality was 36.6 (the higher the coefficient the more inequality) in 2006 compared to its
first measurement of 33.2 in 1995 indicates a rising inequality; there is every indication that
it’s gone up significantly since then. Chile and Botswana that Mongolia sees as role model
economies in terms of its resource management have Gini coefficients of 52.3 and 61
respectively.
According to Guriev, Plekhanov and Sonin (2009) there is an important interaction between
resource wealth and inequality. When resource rents are large, it is easier to buy off the voters
without achieving real redistribution or implementing development-friendly policies.
Conversely, when total rents are appropriated by fewer individuals, rent-seeking (and weak
institutions that make rent-seeking possible) become even more attractive to the members of
the elite. In this way, weak institutions and high inequality can feed off each other in an
economy with large natural resources.
High inequality can be harmful for growth for several reasons. In an unequal society with
imperfect capital markets, many talented people will have no access to capital or education,
resulting in individual poverty traps. High inequality may also bias government policies
towards redistribution policies that hurt growth, as the relatively poor median voter would
prefer to have more redistribution (Persson and Tabellini, 2000, and Acemoglu, Robinson, and
Verdier, 2004), which may result in procyclical fiscal policy (Alessina, Tabellini, and
Campante, 2008).
Figure 29 shows the changes in the income inequality of a group of countries that have similar
socio-economic backgrounds as Mongolia35. As a rough indication, the countries above the red
line are those that have deteriorating income inequality.
Increasing the quality of and access to education, especially the primary education, can
improve the labor force competitiveness and encourage manufacturing that requires higher-
skills. Moreover, there are some policy measures that are outdated and could even exacerbate
income inequality. These include de-facto exemption of dividend and interest income from the
income tax law and that of the residential properties from the real estate tax law.
35 Note that the dates the countries first measured their Gini coefficients are different but the most recent ones are
always in 2006.

ERI Discussion Paper Series No. 2
72
Figure 29: Gini coefficients of the former communist countries of Eastern Europe and Central Asia.