The greening of the accountants

Are we overestimating the value of production in Kazakhstan? 

Most countries regard the value of their production – “gross domestic product” – as the barometer of welfare.  When GDP per capita rises, the government crows of this alleged improvement in a resident’s well-being.

And it’s usually right.  Adjusted for price changes, GDP correlates with life expectancy, education, and the lack of corruption.  While GDP is often a consequence as well as a cause of such factors, most people want more of it.  They’d rather be rich than poor.

Even so, GDP as a barometer has a few cracks in the casing.  Sometimes it will rise although well-being falls.  In World War II, national spending of the belligerents increased, but so did their loss of life. 

The case that dominates headlines today is environmental policy.  When it comes to pollution, GDP is perverse.  Suppose that bronchitis, induced by Almaty’s infamous smog, lands you in the bol’nytsa for a weekend.  Your hospital bill is $300.  GDP will rise by $300, but you certainly are not better off than you would have been without the bronchitis.

The bucks shouldn’t stop here

In the last two or three decades, nongovernmental organizations like the World Resources Institute have estimated “green GDP” by subtracting environmental costs from conventional gross domestic product.  Green GDP has the salubrious effect of penalizing pollution.  If we made it a policy always to measure GDP in green terms, then a developing country seeking to join the ranks of the rich would have to consider whether some pollutant unduly threatened health despite creating a few jobs.

But measuring green GDP is not as clear-cut as it may look.  Let’s return to your case of bronchitis.  We know that the disease doesn’t raise your well-being by $300, so we should subtract at least this amount from GDP.  But the matter doesn’t end there:  We should also penalize production for creating the bronchitis in the first place.  At first, it may seem that we should subtract twice $300, or $600 in all, from GDP.  But that may not be correct.  We do observe that you were willing to pay $300 to avoid future bronchitis; but the cost of the bronchitis that you have already suffered may be less than, equal to, or more than $300.  All that we can say is that the initial subtraction of $300, which is all that some green GDP accountants do, is almost certainly an underestimate of pollution damages.     

A little too green

To make matters worse, sometimes green GDP goes too far.  In Central Asia, some published measures of green GDP assume that all environmental damages fall upon the polluting population. 

For some pollutants, that assumption would be reasonable. For example, when a factory dumps raw sewage into a stagnant creek, it creates algae and bacteria that consume the oxygen dissolved in the water, killing off fish and marine species.  The costs of this pollutant, called “biochemical oxygen demand” or BOD, fall mainly on the local fishermen and fish consumers; it is an example of a “local pollutant.” BOD would not have been so deadly had the creek been fast-moving, since this would have diluted the sewage. 

However, measures of green GDP in Kazakhstan emphasize not BOD but carbon emissions, which result from burning coal and oil in order to generate electricity.  The estimates of carbon damages are so large that they dominate the calculated gap between conventional and green GDP.  So these studies may conclude that a fine way to raise well-being in Kazakhstan is to stop burning fossil fuels.

That conclusion is dead wrong. 

Carbon emissions are the textbook example of a nonlocal pollutant, since their main effects occur far from the point of emission.  Being light, carbon gases rise to the top of the troposphere, where they disperse around the world.  The layer of carbon gases seals in the heat that was radiated by the surface of the Earth -- the “greenhouse effect.”  The atmosphere heats up, warming the surface.  Polar ice melts, and coasts flood.  The damages due to the carbon emissions are global, not local; they fall on Kazakhstan only to the extent that it is part of the world population. Thus, if D denotes the total environmental damages due to emission of a carbon ton, and if T is the number of tons emitted by Kazakhstan, then the local damages due to Kazakhstan’s carbon emissions is not TD, which is what the studies assume, but (crudely) aTD, where a is Kazakhstan’s share of the global population – a tiny fraction. 

For example, suppose that Kazakhstan emits one ton of carbon; that the total damages due to that ton (in themselves a trick to calculate) are estimated at $20.  Kazakhstan holds roughly two-thousandths of the world population, according to World Bank data. Then the resulting damages are Kazakhstanis are worth .002*1*$20 = $.04, or four cents.  The studies, however, would have estimated the damages as $20.

These studies have severely overestimated the impact of carbon emissions on Kazakhstani welfare.  Indeed, Kazakhstan has persisted in emitting carbon gases precisely because it does not incur most of the damages; they fall instead on the rest of the world.

Green GDP accountants are hoisted on their own petard.  Once officials in Astana understand that most carbon damages occur outside Kazakhstan, they may dismiss green GDP as inconsequential.  That would be a grievous error, because production in Kazakhstan does generate local pollutants. 

n        --- Leon Taylor tayloralmaty@gmail.com

n   

Notes

(1)  “Bol’nytsa” is Russian for “hospital.”

(2)  Of course, population is not the only determinant of global warming damages.  Others include income and geography.

(3)   Green GDP should also include the value of environmental services, such as the gorgeous mountain view rimming the south of Almaty.  For simplicity, I will concentrate here on pollution costs and depletion costs.

Good reading

Boyd, James.  The nonmarket benefits of nature:  What should be counted in green GDP?  Resources for the Future Discussion Paper 06-24.  May 2006. http://www.rff.org/files/sharepoint/WorkImages/Download/RFF-DP-06-24.pdf

John Houghton.  Global warming: The complete briefing.  Third edition.  Cambridge University Press.  2004.

Robert N. Stavins, editor.  Economics of the environment:  Selected readings.  Sixth edition.  New York:  W. W. Norton.  2012.

Tietenberg, Tom.  Environmental and natural resource economics.  Seventh edition.  Boston: Pearson Addison Wesley.  2006.    

World Bank.  World Development Indicators.  www.worldbank.org

Conway's experts

Yesterday, the Silk Road Intelligencer reported this:

“Kazakhstan wracked up a $4b current account deficit in the first nine months of 2015, the Central Bank said on Wednesday, a sharp drop from a surplus of over $6b during the same period last year.

“A country’s current account is the difference between government revenue and expenditure….”

Uh, no.  The current account usually refers to the difference between exports and imports.  The account is in deficit when it is negative – that is, when imports exceed exports.  This is the case in Kazakhstan this year because of low oil prices.  For better or worse, Kazakhstan’s economy is based on export revenues from oil and gas, which usually account for at least a fourth of the value of production on Kazakhstani soil (gross domestic product).

The difference between government revenues and spending is called the fiscal surplus.   When spending exceeds revenues, we have a fiscal deficit, which the government covers by borrowing. For instance, if the government receives $1 in tax payments and spends $3, it must borrow $2.

The Central Bank was talking about trade, not fiscal policy.

Unfortunately, English-language coverage of Central Asian economics is abysmal.  This latest gaffe from The Conway Bulletin is just another example.  –Leon Taylor tayloralmaty@gmail.com

Notes

Technically, the current account also includes international income transfers, but these play a minor role in the account.  As the media use the term, the “current account” almost always refers to net exports of goods and services. If you’d prefer a more precise term for net exports, try “the balance of trade.”

Good reading

Paul Krugman and Maurice Obstfeld.  International economics: Theory and policy.  Eighth edition.  Clear and authoritative.

Whither the tenge?

Is the exchange rate headed for 400 tenge per US dollar?

Since its inception in late 1993, the tenge has been devalued sharply by the National Bank of Kazakhstan in April 1999, February 2009, February 2014 and August 2015.  Of these decisions, the most significant by far was this year’s policy to let the foreign exchange market set the exchange rate -- albeit with Bank interventions when the tenge began falling off a cliff in September, with some Almaty kiosks selling a United States dollar for 300 tenge.  Before the float, the exchange rate had been 185 or a bit higher. 

My statistical model finds that the devaluations before 2015 did not permanently affect the change in the exchange rate from one month to the next.  Only the August 2015 decision mattered (Table 1, in the appendix).  The devaluations in 1999, 2009 and 2014 were once-and-for-all increases in the exchange rate, with no impact on the long-run rate of depreciation.  But the August 2015 decision touched off a rise in the exchange rate that continued throughout September.   

The stability, or lack of it, in the exchange rate is due primarily to the National Bank.  Before 2015, the amount of monthly change in the exchange rate (measured in tenge) was extreme only when the Bank devalued; non-Bank shocks were small and fleeting.   Even when the Bank stepped in, the fluctuation in the tenge rapidly disappeared – within one month, for the 2009 and 2014 devaluations; in two or three months, for the 1999 devaluation.  The only time that the exchange rate threatened to become volatile for a long time was when the Bank put it on a float on August 20. 

In my forecast, the August decision to float the tenge may continue to weaken it throughout 2016.  The monthly exchange rate rises at a steady but slowing rate to about 297 in December 2016.  The 95% confidence interval is fairly tight.  For example, in December 2015, the expected exchange rate is 289.06, but other likely values range from 288.1 to 290.02.

Straying the course?

These predictions assume that the new governor of the National Bank, Daniyar Akyshev, continues the exchange-rate policy of his predecessor, Kairat Kelimbetov, who was very publicly sacked on November 3.  Two days after the dismissal, the Bank said it would continue to float the tenge but “reserve[d] the right to smooth large fluctuations that do not reflect supply and demand balance as well as fundamental factors.”  This is what the Bank had been doing under Kelimbetov, and indeed the Bank called its new stance “consistent” with its old policy.  

This declaration of staying the course leads to troubling questions.  First, if the Bank plans to stick to Kelimbetov’s policy, what was the point of firing him? 

Second, if the Bank governor serves only at the pleasure of the President, then can it commit to policies with long-run benefits but short-run costs?  The Bank has had four governors in less than a decade.  The answer is murky because at times political intervention may have helped the economy in the long run.  For example, in 2009, when the tenge was over-valued to the point of a near collapse, the brand-new governor, Gregory Marchenko, devalued it by 25%, which may have ameliorated an ensuing economic slowdown.  But Western doctrine holds that a central bank should be free of short-run politics.

The panic bank

Most troubling of all was the Bank’s rationale for its November 5 policy.  The Bank said it was minimizing intervention in the forex market “in order to preserve its foreign exchange assets of the National Bank and the National Fund.” (The National Fund is the government’s coffers of oil-export tax revenues.)  One may interpret this as an admission that the Bank can no longer defend the tenge, undermining its implicit vow to intervene when the tenge yo-yos.  Also, the Bank postponed indefinitely its monetary policy meeting, which had been scheduled for November 6.  This suggests that the Bank is not sure of what to do.  These admissions of the Bank’s weakness and uncertainty can only lead to depreciation of the tenge. 

How much depreciation?  That’s not clear.  One possibility is that the market on November 5 had anticipated the Bank’s announcement.  In that event, the depreciation that day of 14.2 tenge per dollar is a once-and-for-all leap in the exchange rate.  Another possibility is that the announcement itself creates uncertainty about what the Bank will do in the months to come, so people will rush to sell tenge.  In that event, the equilibrium exchange rate is not clear, but the most conspicuous candidate is 400 tenge per dollar – just as 300 tenge per dollar in recent weeks had provided a focal point.  As I write this – Monday, November 9 – the immediate exchange rate has jumped to 312.65 tenge, so I think that an eventual monthly rate of 400 tenge is not impossible.

A moderate approach assumes that the November 5 event was a once-and-for-all depreciation that took until November 9 to fully unfold.  This amounts to a onetime increase in the monthly exchange rate of 28-29 tenge, a rise of more than 10%.  The exchange rate is headed for a medium-run equilibrium of 326-327 by the spring of 2017.    

In the forex market for the tenge, conditions are ideal for the perfect storm.  The National Bank says it will stick to a policy that the government has denounced as a failure, because the Bank doesn’t know what else to do.  Bank officials also say they don’t know when they will meet in order to discuss a new plan.  Thus the Bank has turned over the forex market to speculators.  Although the Bank says it “reserves the right” to intervene to stop sharp fluctuations of the exchange rate, it says in the same breath that it can’t afford routine interventions.  How then can it afford a major one?

The Bank’s confused policy signals to speculators the opportunity for a killing.  Since late August, the tenge has become one of the world’s most volatile currencies.  The market is thin, and a large short sale of tenge can appreciably affect the exchange rate – particularly because the Bank, which had recently accounted for 60% of the volume of transactions, has said that it has pulled out of the market.  – Leon Taylor, tayloralmaty@gmail.com

Notes

All tenge statistics are from the National Bank of Kazakhstan on its Web page www.nationalbank.kz .

     

Appendix

Dependent Variable: DTENGE
Method: Least Squares
Date: 11/03/15   Time: 18:43
Sample (adjusted): 1995M06 2015M09
Included observations: 244 after adjustments
Variable	Coefficient	Std. Error	t-Statistic	Prob.
DTENGE(-1)	0.488	0.113	4.302	0.000
DTENGE(-18)	0.170	0.060	2.833	0.005
NEWDAUG15	44.828	3.507	12.782	0.000
NEWDFEB09	-5.286	3.969	-1.332	0.184
NEWDFEB14	-0.061	3.585	-0.017	0.987
NEWDAPRIL99	-4.399	4.012	-1.096	0.274
R-squared	0.615	Mean dependent var		0.799
Adjusted R-squared	0.607	S.D. dependent var		4.650
S.E. of regression	2.917	Akaike info criterion		5.003
Sum squared resid	2024.612	Schwarz criterion		5.089
Log likelihood	-604.369	Hannan-Quinn criter.		5.038

Table 1:  Regresses the first difference of the exchange rate on its first and 18^th lags as well as on first differences of the four intervention dummies.

In Table 1, DTENGE(-k) denotes the kth lag of the first difference of the monthly exchange rate, which is the dependent variable.  Both lags are statistically significant.  I don’t know why the 18^th lag matters, but evidently it is not because of seasonality, since the simple and partial autocorrelation functions indicated that no other lags beyond the first affected the current first difference by much.

The other four independent variables in Table 1 are first differences of the dummies for interventions by the National Bank of Kazakhstan in the foreign exchange market.  The interventions – all of them devaluations -- occurred in April 1999, during the ruble crisis; February 2009, during the global financial crisis; February 2014, after depreciation of the ruble; and August 2015, when annual oil prices were plunging on the spot market.  Of the four interventions, only the most recent one affected the month-to-month increase in the exchange rate with statistical significance. And its practical significance, measured by its coefficient, is quadruple that of the other three interventions.

Table 2 below is the model used for the forecasts.

Dependent Variable: DTENGE
Method: Least Squares
Date: 11/03/15   Time: 18:47
Sample (adjusted): 1995M06 2015M09
Included observations: 244 after adjustments
Variable	Coefficient	Std. Error	t-Statistic	Prob.
DTENGE(-1)	0.379	0.063	6.038	0.000
DTENGE(-18)	0.176	0.060	2.955	0.003
NEWDAUG15	46.591	3.143	14.822	0.000
R-squared	0.610	Mean dependent var		0.799
Adjusted R-squared	0.607	S.D. dependent var		4.650
S.E. of regression	2.914	Akaike info criterion		4.989
Sum squared resid	2046.596	Schwarz criterion		5.032
Log likelihood	-605.686	Hannan-Quinn criter.		5.007

Table 2: Regresses the first difference of the exchange rate on its own first and eighteenth lags as well as on the first difference of an intervention dummy. 

Date: 10/31/15   Time: 20:04
Sample: 1993M11 2015M09
Included observations: 263
Autocorrelation	Partial Correlation		AC	PAC	Q-Stat	Prob
.|*******	.|*******	1	0.956	0.956	242.91	0.000
.|*******	.|*     |	2	0.924	0.123	470.80	0.000
.|******|	.|.     |	3	0.896	0.054	686.19	0.000
.|******|	.|.     |	4	0.869	0.007	889.55	0.000
.|******|	.|.     |	5	0.843	0.003	1081.6	0.000
.|******|	.|.     |	6	0.818	0.006	1263.1	0.000
.|******|	.|.     |	7	0.796	0.026	1435.6	0.000
.|******|	.|.     |	8	0.775	0.015	1599.8	0.000
.|***** |	.|.     |	9	0.755	0.006	1756.1	0.000
.|***** |	.|.     |	10	0.735	-0.000	1904.9	0.000
.|***** |	.|.     |	11	0.715	-0.003	2046.3	0.000
.|***** |	.|.     |	12	0.695	-0.009	2180.5	0.000
.|***** |	.|.     |	13	0.675	-0.005	2307.7	0.000
.|***** |	.|.     |	14	0.657	0.001	2428.4	0.000
.|***** |	.|.     |	15	0.638	-0.006	2542.7	0.000
.|****  |	.|.     |	16	0.620	-0.000	2651.0	0.000
.|****  |	.|.     |	17	0.602	0.001	2753.6	0.000
.|****  |	.|.     |	18	0.585	-0.004	2850.9	0.000
.|****  |	.|.     |	19	0.565	-0.030	2942.2	0.000
.|****  |	.|.     |	20	0.546	-0.014	3027.7	0.000
.|****  |	.|.     |	21	0.532	0.044	3109.2	0.000
.|****  |	.|.     |	22	0.516	-0.010	3186.3	0.000
.|****  |	.|.     |	23	0.502	0.009	3259.3	0.000
.|****  |	.|.     |	24	0.487	-0.003	3328.6	0.000
.|***   |	.|.     |	25	0.473	0.002	3394.2	0.000
.|***   |	.|.     |	26	0.460	-0.002	3456.4	0.000
.|***   |	.|.     |	27	0.446	-0.005	3515.1	0.000
.|***   |	.|.     |	28	0.432	-0.004	3570.6	0.000
.|***   |	.|.     |	29	0.419	-0.010	3622.8	0.000
.|***   |	.|.     |	30	0.404	-0.015	3671.6	0.000
.|***   |	.|.     |	31	0.389	-0.016	3717.0	0.000
.|***   |	.|.     |	32	0.374	-0.014	3759.2	0.000
.|***   |	.|.     |	33	0.358	-0.014	3798.1	0.000
.|**    |	.|.     |	34	0.343	-0.009	3834.0	0.000
.|**    |	.|.     |	35	0.328	-0.011	3866.9	0.000
.|**    |	.|.     |	36	0.313	-0.009	3897.1	0.000

Figure 1: Autocorrelation functions for the monthly rate of exchange of tenge for a US dollar, November 1993 through September 2015. 

The last two figures concern how I selected the forecast model.  Figure 1 suggests that an autoregressive function of the exchange rate with one or two lags might be stationary.  No seasonal effect is evident.  The simple autocorrelation function dies down slowly, and the partial autocorrelation function cuts off after one or two lags.

Date: 10/31/15   Time: 20:18
Sample: 1993M11 2015M09
Included observations: 262
Autocorrelation	Partial Correlation		AC	PAC	Q-Stat	Prob
.|**    |	.|**    |	1	0.332	0.332	29.245	0.000
.|*     |	.|.     |	2	0.090	-0.023	31.388	0.000
.|.     |	.|.     |	3	0.029	0.007	31.607	0.000
.|.     |	.|.     |	4	0.027	0.019	31.797	0.000
.|.     |	.|.     |	5	0.025	0.011	31.963	0.000
.|.     |	.|.     |	6	0.026	0.015	32.148	0.000
.|.     |	.|.     |	7	0.019	0.006	32.248	0.000
.|.     |	.|.     |	8	0.023	0.015	32.386	0.000
.|.     |	.|.     |	9	0.019	0.006	32.481	0.000
.|.     |	.|.     |	10	0.007	-0.003	32.497	0.000
.|.     |	.|.     |	11	0.010	0.008	32.526	0.001
.|.     |	.|.     |	12	-0.005	-0.014	32.534	0.001
.|.     |	.|.     |	13	-0.036	-0.035	32.892	0.002
.|.     |	.|.     |	14	-0.016	0.007	32.967	0.003
.|.     |	.|.     |	15	-0.007	-0.002	32.979	0.005
.|.     |	.|.     |	16	-0.010	-0.008	33.006	0.007
.|.     |	.|.     |	17	0.006	0.014	33.015	0.011
.|*     |	.|*     |	18	0.143	0.157	38.847	0.003
.|*     |	.|*     |	19	0.191	0.111	49.257	0.000
.|.     |	*|.     |	20	0.026	-0.090	49.458	0.000
.|.     |	.|.     |	21	-0.007	0.004	49.471	0.000
.|.     |	.|.     |	22	-0.029	-0.030	49.707	0.001
.|.     |	.|.     |	23	-0.012	-0.002	49.750	0.001
.|.     |	.|.     |	24	-0.014	-0.017	49.805	0.001
.|.     |	.|.     |	25	-0.012	-0.010	49.847	0.002
.|.     |	.|.     |	26	0.011	0.017	49.882	0.003
.|.     |	.|.     |	27	0.011	-0.002	49.920	0.005
.|.     |	.|.     |	28	0.023	0.022	50.073	0.006
.|.     |	.|.     |	29	0.032	0.022	50.376	0.008
.|.     |	.|.     |	30	0.039	0.024	50.839	0.010
.|.     |	.|.     |	31	0.006	-0.004	50.850	0.014
.|.     |	.|.     |	32	-0.004	0.006	50.856	0.018
.|.     |	.|.     |	33	-0.020	-0.026	50.974	0.024
.|.     |	.|.     |	34	-0.010	0.002	51.004	0.031
.|.     |	.|.     |	35	-0.006	0.001	51.013	0.039
.|.     |	.|.     |	36	-0.014	-0.031	51.075	0.049

 Figure 2:  Autocorrelation functions for the first difference of the monthly rate of exchange of tenge for a US dollar.

In Figure 2, the simple autocorrelation function for the monthly change in the exchange rate dies down, or cuts off, after one lag.  The partial autocorrelation function cuts off after one lag.  Weak correlations occur at the eighteenth and nineteenth lags.  No seasonal effect is evident.  The patterns suggest that the first difference of the exchange rate is an autoregressive function of the first, eighteenth and nineteenth lags.  In my OLS model, I dropped the nineteenth lag because of statistical insignificance.

Dataset:  November 1993 through September 2015 

Month	Tenge
1	4.69
2	5.82
3	8.17
4	11.41
5	16.70
6	23.59
7	35.60
8	41.73
9	44.66
10	45.69
11	47.04
12	48.62
13	51.02
14	53.47
15	55.43
16	58.67
17	60.49
18	62.09
19	63.11
20	63.54
21	62.62
22	56.69
23	59.91
24	61.54
25	63.35
26	63.97
27	64.30
28	65.20
29	65.22
30	65.50
31	66.44
32	66.80
33	67.03
34	67.34
35	68.14
36	69.18
37	69.96
38	72.54
39	74.70
40	75.63
41	75.44
42	75.24
43	75.46
44	75.50
45	75.53
46	75.55
47	75.55
48	75.55
49	75.55
50	75.55
51	76.09
52	76.40
53	76.44
54	76.50
55	76.58
56	76.75
57	77.18
58	77.83
59	79.39
60	80.96
61	82.21
62	83.31
63	84.40
64	85.18
65	86.75
66	111.05
67	118.50
68	130.36
69	132.20
70	131.95
71	135.21
72	140.86
73	139.63
74	138.22
75	139.02
76	139.77
77	141.25
78	142.17
79	142.30
80	142.50
81	142.70
82	142.66
83	142.72
84	142.64
85	143.56
86	144.31
87	145.09
88	145.23
89	145.42
90	145.52
91	145.95
92	146.40
93	146.69
94	147.06
95	147.52
96	147.93
97	148.43
98	149.59
99	151.14
100	151.76
101	152.12
102	152.54
103	152.90
104	153.10
105	153.52
106	154.07
107	154.42
108	154.40
109	154.30
110	155.08
111	155.53
112	153.98
113	151.55
114	151.82
115	151.21
116	149.15
117	146.94
118	146.76
119	147.90
120	147.92
121	147.07
122	145.08
123	141.20
124	139.18
125	139.01
126	138.20
127	137.12
128	136.38
129	135.56
130	136.16
131	135.43
132	133.26
133	130.75
134	130.04
135	130.11
136	130.13
137	130.53
138	131.37
139	131.37
140	133.75
141	135.66
142	135.52
143	134.31
144	133.83
145	134.10
146	133.88
147	133.13
148	131.40
149	128.76
150	126.94
151	122.62
152	119.76
153	118.13
154	122.63
155	126.20
156	127.66
157	127.92
158	127.93
159	125.74
160	124.79
161	124.03
162	122.19
163	120.23
164	121.96
165	122.09
166	124.85
167	122.46
168	120.84
169	120.69
170	120.78
171	120.35
172	120.34
173	120.67
174	120.50
175	120.56
176	120.70
177	120.29
178	120.02
179	119.67
180	119.85
181	120.06
182	120.58
183	121.27
184	144.90
185	150.73
186	150.71
187	150.34
188	150.34
189	150.62
190	150.78
191	150.87
192	150.79
193	149.92
194	148.69
195	148.09
196	147.87
197	147.14
198	146.72
199	146.67
200	147.05
201	147.51
202	147.35
203	147.37
204	147.58
205	147.50
206	147.41
207	147.05
208	146.45
209	145.76
210	145.45
211	145.56
212	145.77
213	145.90
214	146.56
215	147.21
216	147.99
217	147.85
218	147.90
219	148.38
220	148.26
221	147.79
222	147.79
223	147.89
224	148.86
225	149.74
226	149.54
227	149.77
228	150.39
229	150.52
230	150.42
231	150.73
232	150.51
233	150.73
234	150.96
235	151.00
236	151.43
237	152.58
238	152.93
239	153.24
240	153.99
241	153.41
242	154.04
243	154.96
244	173.36
245	182.31
246	182.04
247	182.42
248	183.51
249	183.52
250	182.07
251	181.96
252	181.47
253	180.87
254	181.81
255	183.70
256	184.92
257	185.31
258	185.73
259	185.80
260	186.04
261	186.80
262	203.62
263	258.17

References

Bruce L. Bowerman, Richard T. O’Connell, and Anne B. Koehler.  Forecasting, time series, and regression: An applied approach.  Fourth edition.  Australia: Brooks/Cole. 2005.

National Bank of Kazakhstan. Press-release No. 58: National Bank reduces its participation in the domestic foreign exchange market to preserve its   reserves.  www.nationalbank.kz. November 5, 2015. 

Robert S. Pindyck and Daniel L. Rubinfeld.  Econometric models and economic forecasts.  Fourth edition.  Boston: Irwin/McGraw-Hill. 1997.