Content uploaded by Sergey Alexeev
Author content
All content in this area was uploaded by Sergey Alexeev on Dec 20, 2021
Content may be subject to copyright.
Democracy does improve human capital
Sergey Alexeev
University of New South Wales, NSW, Australia
November, 2021
s.alexeev@unsw.edu.au
Abstract Smart Voting (SM) is an online service designed by the Russian po-
litical opposition to promote candidates with the best chance to defeat Kremlin
approved politicians. In this paper, using annual county-level data on public
expenditure and a triple difference (TD) strategy, I utilize a plausibly exogenous
increase in political competition caused by an unexpected success of SM to show
that democracy promotes public expenditure on health and education. This
finding establishes the empirical linkage between democracy and human capital.
This linkage has been hard to establish so far.
JEL codes: D72; I10; O15; O17
Key words: Voting Behavior; Health; Human Resources; Institutional Arrangements
Declarations of interest: none
This research did not receive any specific grant from funding agencies in the public, commercial,
or not-for-profit sectors.
The funding sources had no involvement in the conduction of the research and/or preparation
of the article; in the collection, analysis and interpretation of data; in the writing of the report;
and in the decision to submit the article for publication.
1 Introduction
The evidence that democracy promotes human capital is neither strong nor ro-
bust. Some estimates suggest that democracies spend more on healthcare and
achieve better health outcomes (e.g., Baum and Lake 2003; Besley and Ku-
damatsu 2006). At the same time, some countries under authoritarian rule
showed the most dramatic improvements in human development (e.g., East Asian
NICs and Communist countries). In contrast, many democracies witnessed and
are still experiencing extreme disparities in their living conditions (e.g., India,
Sub-Saharan Africa, Latin America). Therefore, many empirical studies conclude
that democracies have little or no impact on components of human development
(Gerring, Thacker, and Alfaro 2012; Ross 2006). These ambiguous findings re-
sult from the complexity of casually estimating the effect of democracy on human
capital. In this paper, I take advantage of unique features of Russian politics
to show that democracy, as measured by an increase in political competition,
increase public expenditure on health and education.
Soon after coming into power in 2000, Vladimir Putin initiated so-called ‘Re-
vision number Six’ (Gatov 2016; Kommersant 2010). The document declares
that the Presidential Administration (PA) – initially an office of the head of
state – becomes an institution responsible for minimizing political freedom in
Russian, assuming that society is not mature enough and can not self-regulate.
This approach and the school of thought originated from the late Soviet counter-
intelligence community and was initially attempted by Andropov, who, similarly
to Putin, was the head of Soviet secret police and later the head of the Union
(Pringle 2000,2001).
2 METHODS AND DATA 2
According to this approach, the PA’s objectives are to directly or indirectly
control political parties, community and political leaders, candidates for elective
positions, election officials, mass media, journalists and civil society organiza-
tions. The revision’s ultimate goal was to build a state where democratic in-
stitutions exist nominally, but in reality, these institutions are fully controlled
by PA. A game-theoretical treatment of this state building approach is analyzed
in Guriev and Treisman (2020). As the ‘Revision number Six’ puts it, this ap-
proach aims to allow predicting and, if required, manufacturing political climates
in Russia and neighbouring countries, allowing the President to perform his ‘du-
ties.’
To ensure that only pro-Kremlin candidates are elected on all levels of gov-
ernment, the opposing candidates with significant support are not allowed reg-
istration. In contrast, the candidates with minor public backing are allowed to
run but only to maintain a public impression that PA backed candidates have
overwhelming support. If situations get out of control, PA instructs the electoral
commission to rig the elections. If rigging the election fails, PA instructs police
or secret service to poison non-compliant politician or fabricate a criminal case
(e.g., Ross 2018).
In September 2018, to break the monopoly of pro-Kremlin politicians, Alexei
Navalny – the most prominent Russian opposition leader – launched the SM
system, an online list of registered opposing candidates that are most likely
to defeat the Kremlin-backed politicians. The aim is to consolidate the votes
and disrupt the Kremlin’s political controls (Turchenko and Golosov 2021). SM
achieved unexpected success in regional elections in September 2020 when pro-
Kremlin politicians lost their majority in legislatures in three Russian regions
of Novosibirsk, Tambov, and Tomsk. The changes in public expenses in those
regions following this unexpected increase in political competition is at the core
of my identification strategy, which I now explain.
2 Methods and data
The goal is to understand how an unexpected emergence of competitive legis-
lature influences the expenditure on public education in health. The ideal ex-
periment for causal identification would be to randomly assign parliaments with
different degrees of competition to territories and observe the subsequent impacts
on public health and education expenditure. To approximate this experiment, I
use an unexpected success of SM in the Russian regions of Novosibirsk, Tambov,
and Tomsk (total of 899 counties).
In particular, I use a TD strategy over time, across regions where politicians
supported by SM succeed to be election and fails, and between public expenditure
on human capital and the rest of spending.
𝑌𝑖𝑡𝑐 =𝛽1{𝑆𝑚𝑎𝑟𝑡𝑐×𝐴𝑓𝑡𝑒𝑟𝑡×𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖}+𝛼2{𝐴𝑓 𝑡𝑒𝑟𝑡×𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖}
+𝛼3{𝐴𝑓𝑡𝑒𝑟𝑡×𝑆𝑚𝑎𝑟𝑡𝑐}+𝛼4𝐴𝑓𝑡𝑒𝑟𝑡+𝛿𝑖𝑐𝑇 𝑟𝑒𝑛𝑑𝑡+𝛾𝑖𝑐 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑡+𝜀𝑖𝑡𝑐
(1)
where 𝑌𝑖𝑡𝑐 is the natural logarithm of public expenditure of type 𝑖, in year 𝑡, in
3 RESULTS AND CONCLUSIONS 3
county (municipality) 𝑐.𝐴𝑓𝑡𝑒𝑟𝑡us an indicator for 2021 onward (after the elec-
tions). 𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖is an indicator for weather expenditure 𝑖is on public health or
education. 𝑆𝑚𝑎𝑟𝑡𝑐is an indicator for whether SM succeed in county 𝑖.𝛿𝑖𝑐 𝑇 𝑟𝑒𝑛𝑑𝑡
controls for differential linear time trends by the full interaction of public expen-
diture types and county fixed effects. 𝛾𝑖𝑐 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑡controls for the interaction of
year, public expenditure types, and country fixed effects.
The identifying assumption is that without SM, the difference between public
expenditures into human capital and other expenditures would have changed
in the same way from the pre-period to the post-period in the counties with
successful and unsuccessful SM (after controlling for public expenditures type
and county-specific seasonality and for differential linear time trends). For an
omitted variable to explain the results, it would have to have a non-linear change
after 2021 that affects the public expenditures into human capital more than it
affects other public expenditures among counties where the SM achieved success,
as compared to counties where it was unsuccessful. The standard errors are
clustered at the region levels because that is where the SM varies.
I apply the model to the official municipality-level statistical data covering
2018-2021. The data is published by the Russian Federal State Statistics Service
and is publicly available online. I utilize the Centre for Advanced Governance
version of this data. The centre scrapes the official data from the website and
structures it as a ready-to-use annual municipality-level panel (Rosstat 2021).1
The expenditures on education and health and the rest can be unambiguously
identified in this data. The information on the regions where SM achieved success
is available in the media.
3 Results and conclusions
Column (1) of Table 1uses data only on health and education to show a difference-
in-differences estimate over time and between regions where parliament became
competitive and stayed uncompetitive. Column (2) uses all types of public ex-
penses and shows a difference-in-differences estimate over time and between hu-
man capital and other expenses. While the two columns use different sources
of variation, they both find statistically significant effects of 5.5% and 3.3%,
respectively.
Column (3) estimates the effect from Column (2) separately for regions where
SM succeeded and failed and shows that the effect is driven by the former. These
regions had an effect of 7.7%, while the point estimate is only 1.1% among regions
where SM failed. The latter effect can be interpreted as a placebo estimate under
the assumption that SM has no effect in regions where SM fails.
Finally, Column (4) shows the results from the main TD specification de-
scribed in Equation (1). I find an effect of 7.7%, statistically significant at the
10% level. In this specification, regions where SM failed serve as a control group.
In Columns (3) and (4) of Equation (1), the coefficient on After×Smart can
be interpreted as a difference-in-differences estimate of the effect of SM on non-
1Data for 2020 and 2021 was taken from the Federal State Statistics Service official website.
Data for 2021 can be requested by contacting the service and will be publicly available in mid-
2022.
3 RESULTS AND CONCLUSIONS 4
Table 1: The effect of the increase in political competition
ln(expenditure)
(1) (2) (3) (4)
After×Smart 0.055** 0.007 0.009
(0.023) (0.014) (0.014)
After×Capital 0.033** 0.023
(0.015) (0.021)
After×Capital×Smart 0.077*** 0.061**
(0.027) (0.031)
After×Capital×Non-Smart 0.011
(0.019)
County×Type×Lin. Tr X X X X
County×Type×Year X X X X
After X X X X
Data type Capital All All All
Observations 81,212 162,424 162,424 162,424
Clusters 85 85 85 85
Notes: This table shows the effect of political competition on public expen-
diture in health and education. The outcome variable is the amount of expen-
diture in a specific category, in a county, and in a year. Column (1) analyzes
only expenditure in health and education using a difference-in-differences
estimate over time and across counties. Columns (2)-(4) include data on
all public expenditures. Column (2) presents a difference-in-differences es-
timate over time and between expenditure types. Column (3) presents the
estimate from Column (2) separately for regions where SM succeeded and
failed. Column (4) presents our primary triple-difference estimate over time,
across counties, and between expenditure types. A county is categorized as
having increased competition if SM prevented the pro-Kremlin party from
securing the majority in the regional parliament in September 2020. County-
level data from 2018 to 2021 is used. Standard errors clustered at the region
level in parenthesis.
*𝑝 < 0.05, ** 𝑝 < 0.01, *** 𝑝 < 0.001.
Source: Rosstat (2021)
human capital expenditure, using variation over time and between regions. Since
I do not expect the effect, this coefficient can be interpreted as a placebo test.
An unexpected establishment of competitive parliament increases public in-
vestment into education and health and, thus, fosters human capital accumula-
tion. This makes two contributions. General contribution is that it shows that
democracy, understood as a competitive market of politicians, improves human
capital.
Another contribution is to demonstrate once again the costs of social experi-
mentation that is happening in Russia (cf. Zhuravskaya, Guriev, and Markevich
2021). The system that Vladimir Putin and his old-boy network created for un-
clear reasons hurt Russian through multiple channels. Apart from the obvious
cost of political repression and uncertain investment climate, my paper shows
that uncompetitive decorative parliament reduces human capital. On a positive
note, my paper suggests that returning to competitive parliament will increase
spending on human capital, reducing funds available for repression and public
opinion manipulation.
REFERENCES 5
References
Baum, M. A. and D. A. Lake (2003). “The Political Economy of Growth: Democ-
racy and Human Capital”. In: American Journal of Political Science 47.2,
pp. 333–347. issn: 00925853, 15405907. url:http://www.jstor.org/stable/
3186142 (cit. on p. 1).
Besley, T. and M. Kudamatsu (May 2006). “Health and Democracy”. In: Ameri-
can Economic Review 96.2, pp. 313–318. doi:10.1257/000282806777212053.
url:https:/ / www.aeaweb. org/articles?id=10 . 1257 /000282806777212053
(cit. on p. 1).
Gatov, V. (2016). No Canaries in the Coal Mine: The Demise of RBC.https://
therussianreader.com/2016/05/15/rbc-russian-free-press-demise/. Accessed:
2021-12-08 (cit. on p. 1).
Gerring, J., S. C. Thacker, and R. Alfaro (2012). “Democracy and Human Devel-
opment”. In: The Journal of Politics 74.1, pp. 1–17. doi:10.1017/S0022381611001113.
eprint: https://doi.org/10.1017/S0022381611001113.url:https://doi.org/
10.1017/S0022381611001113 (cit. on p. 1).
Guriev, S. and D. Treisman (2020). “A theory of informational autocracy”. In:
Journal of Public Economics 186, p. 104158. issn: 0047-2727. doi:https:
//doi.org/10.1016/j.jpubeco.2020.104158.url:https://www.sciencedirect.
com/science/article/pii/S0047272720300220 (cit. on p. 2).
Kommersant (2010). Revision Number Six.https://www.kommersant.ru/doc/
1385567. Accessed: 2021-12-08 (cit. on p. 1).
Pringle, R. W. (2000). “Andropov’s Counterintelligence State”. In: International
Journal of Intelligence and CounterIntelligence 13.2, pp. 193–203. doi:10.
1080/08850600050129718. eprint: https://doi.org/10.1080/08850600050129718.
url:https://doi.org/10.1080/08850600050129718 (cit. on p. 1).
— (2001). “Putin: The New Andropov?” In: International Journal of Intelligence
and CounterIntelligence 14.4, pp. 545–558. doi:10.1080/08850600152617155.
eprint: https://doi.org/10.1080/08850600152617155.url:https://doi.org/
10.1080/08850600152617155 (cit. on p. 1).
Ross, C. (June 2018). “Regional elections in Russia: instruments of authoritarian
legitimacy or instability?” In: Palgrave Communications 4.1, p. 75. issn: 2055-
1045. doi:10.1057/s41599-018-0137-1.url:https://doi.org/10.1057/s41599-
018-0137-1 (cit. on p. 2).
Ross, M. (2006). “Is Democracy Good for the Poor?” In: American Journal of
Political Science 50.4, pp. 860–874. doi:https:// doi.org/ 10.1111/ j.1540-
5907.2006.00220.x. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/
j.1540-5907.2006.00220.x.url:https://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1540-5907.2006.00220.x (cit. on p. 1).
Rosstat (2021). Database of municipalities. processed by Research Data Infras-
tructure (Center for Advanced Governance). url:http://www.data-in.ru/
data-catalog/datasets/115/ (cit. on pp. 3,4).
Turchenko, M. and G. V. Golosov (2021). “Smart enough to make a difference?
An empirical test of the efficacy of strategic voting in Russia’s authoritarian
elections”. In: Post-Soviet Affairs 37.1, pp. 65–79. doi:10.1080/1060586X.
2020.1796386. eprint: https://doi.org/10.1080/1060586X.2020.1796386.url:
https://doi.org/10.1080/1060586X.2020.1796386 (cit. on p. 2).
