JournalDashmir Asani, IJSRMAcademic Publisher10.18535/ijsrm/v6i9.em02Some Remarks on the Estimation of Informal Economy in the Republic of MacedoniaDashmir AsaniDode PrengaElmira KushtaDepartment of Mathematics, Faculty of Technical Sciences, University “Ismail Qemali”AlbaniaDepartment of Physics, Faculty of Natural Sciences, University of TiranaAlbaniaPhD student, Department of Informatics and Statistics, Faculty of Economy, University of TiranaAlbania20180609715
In this paper we discuss some findings on the estimation of the informal economy in a particular economical system, the Republic of Macedonia. We observed that undeviating application of standard models were likely to produce inconsistent results and therefore some preliminary analytic precaution have to been taken. We obtained that the set of variables that mostly influenced the informality and shadow economy could be assigned by a step by step selecting procedure, instead of straight CDA or MIMIC model application. Next we assumed that the nature of influence for each variable could be proofed and not implemented by general arguments. Finally the calculations were performed for time intervals when the overall state appeared to be more relaxed. By performing in this way a step by step analysis we concluded in acceptable values and in the evidencing of dominant factors that simulate informal economy for the country. Not surprisingly a non-model variable-remittances-has been evidenced as an important factor. The work reveals the importance of model's basic assumption fulfilling in the case of economic systems where observables are non-stationary and the economy itself is typically in dynamical evolution. As result of such improvement of the procedure we obtained that the country's informal economy is going slight down starting from a local peak value touched in 2010-2011. By now it is stabilized around the values 33%-35% of GDP. We obtained herein that the majority of economic and political measures undertaken recently in the framework of informality reduction have worked in the aimed direction. From the factor point of view we obtained that contribution of some variables in the informal economy growth have been very specific and even contrarily to the expatiation for the period analyzed, that highlighted the importance of direct and continues calculation rather than judging from general perspective approaches.
Informal and shadow economycurrency demand approachtaxation policyMIMIC model1. Introduction
Informal economy is defined as part of a country’s
economy that is not observable in the sense of
state fiscal activities. In a more detailed discussion
[ [1] ], [ [2] ] this definition is reserved only for the part
of unregistered economy that is created as result
of evading formal official records to simply
escape the Taxes’ duties. There is another part of
unregistered economy that could originate from
criminal activities [ [3] ] which is expected to have
different structure. A more general concept for
those unregistered economical activates is
encapsulated in the definition “shadow or
underground economy”. In the analysis of the
informal economy there are two major objectives:
estimation of the size and identification of factors
that cause it to exits. In this view, researchers have
evidenced the fact that the un-registered is
tractable however, and therefore it can be
estimated making use some theoretical
relationship with its indicators. Moreover, factors
and causes of informality can be analyzed in this
process too. In econometrics it is believed that the
existence of informal economy causes
discrepancies, distortions for many economic
parameters as compared to their expected values.
So, it might cause a part unemployment force to
be concealed, can cause the consumption to be
higher than production or the money in circulation
to be larger than expected. Accordingly this
unseen economic quantity has been modeled and
estimated [ [1] ], [ [2] ] etc., by simple regression
techniques.. Direct and indirect methods have
been proposed but each one has its advantages and
disadvantages.[ [1] ], [ [4] ] Practical evidences show
that obtaining an appropriate model and fixing the
set for variables in each one of them depends on
concrete economy under study as seen ne [ [5] ], [ [6] ],
[ [7] ] an many others considerations. If data records
fulfill some necessary requirements, the first
evidence and the easiest way to estimate the
informal economy is considered the method of
“the GDP discrepancy”. It states that the
difference between GDP calculated by
consumption and incomes give the informal
economy
fig-68e75173f8fc8dfbcab270a5b80d4178
GDPInformal GDPExpenditures GDPIncomes (1)
In the reference[ [1] ] it is highlighted that results of
the equation (1) would be admissible if data used
in it would have been recorded carefully and
without any subjective distortion. Therein is
noticed that usually it didn’t happen, therefore
when using (1) the error should not be neglected.
Other discrepancies methods based on physical
outputs do need again adequate databases.
Nevertheless, some of those methods are
complicated and their techniques of calculation
seem to be questionable in some critical point of
view. But in any sense, discrepancies have been
largely considered as indirect indicators and traces
of the Informal economy and therefore they have
been acknowledged as the basis of direct
calculation of the informal economy. In this
category have been listed discrepancy of
unemployment rate, electricity use etc.[ [1] ],[ [3] ] etc.
Aside those direct methods, some indirect
techniques have been found more useful. So using
simple currency ratio (SCR) by simple formula.[ [2] ]
C f D
fig-cd0130ade1b2bdf3583d25dd6d2c37fd
(2)
where C and D are respectively the currency in
circulation and the Deposits, indices i and f stand
for “informal” and “formal” whereas The
calculation according (2) is straightforward and is
detailed in[ [1] ] etc. The confidence of the
estimation (2) depends on two elements: the
stationary of the series which we discussed
especially in [ [8] ] and the rigorousness of the
assumption that in actual economy velocities of
the money in formal and informal sector are equal.
Moreover the problems of zero informality time as
by direct modeling in[ [1] ] would mostly affect the
calculation in the case of transitory economies.
Next, another technique called the currency
demand approach (CDA) with following formulae
or its derivative have been used largely:
in
fig-afdbb35c06d0f2c6e2f6bf0556790ace
(3)
Here M is narrow/broad money, C the money in
circulation, Tax is the average tax rate and R is the
interest rate for Deposits. Again informal
economy is supposed to be proportional to the
extended demand for money and therefore the
assumption of equal money velocity is taken as
true. In applications (1), (2) and other similar, the
informal economy is modeled as linear form (or
log linear) of some factors (cause variables).
Generally speaking, if a candidate-factor included
in the forms (1), (2) etc., has a good regression
statistics, one agrees that it is a cause for informal
economy. A more complex method but probability
more effective is the MIMIC model that put the
informal economy in the middle point between
factors and indicators. It looks like following
fig-b1634dbb8ef1e145e5054b823ede474b
(4)
where in IndicatorY could be placed any
macroeconomic index areas in FactorX could be
whichever economic parameter or variable. In
general, factors are parameters like Taxes, tariffs,
the incomes, wages, contributions, a quantifier of
the structure of the capital for the country,
political performance, and many others. Informal
or shadow economy plays the role of the
intermediate transcript between sides of relation
(4), starting as response variable for the set of
factors, and becoming factor variable for final
indicators. Therefore a double regression is
needed in this case which inevitably needs for
more careful procedures as have been developed
in [ [9] ], [ [10] ] etc. Apparently the mathematical nature
of the series of variables is expected to affect the
estimation. This aspect is taken into consideration
in our calculation. Finally, another source of
incertitude in the evaluation of the latent variable
under examination is related to the real structure
of the unregistered economy. So, the presence of
criminal or similar activities that contribute in the
shadow economy is inevitable and this last cannot
be modeled so the overall calculation became
complicated [ [11] ]. As a result, by direct
implementation of the standard modeling it is
arguable that values obtained could not be always
satisfactory or even realistic. Based on those
remarks, we have anticipated the applications of
linear models mentioned herein by analyzing each
element discussed in this introduction.
2. Some findings using direct and indirect methods
The economy of Republic of Macedonia has
known an energetic change at 1990 when the
system has been transformed toward the market
economy. In this sense it is expected that the
variables representing economic observables have
been highly dynamical during the afterwards
period. We commented this feature in our recent
work.[ [8] ] We acknowledge that during [1990,
2016], methodical improvement in public
database have been applied consecutively until a
full modernized methodology has been adopted by
the end of 2010.[ [12] ],[ [13] ] In this view we expected
that linear modeling of the informal economy
might have considerable incertitude or other
incompatibilities due the lack of fulfilling of some
assumptions. In our first calculation we observed
that the literally application of standard models
leads to different results. From the other side, it
happen that an observable that does not appear as
factor in the regressions of the type (3) or (4), has
been identified as related with informal economy
calculated using another model. Hence we
addressed them in the followings.
2.1. Avoiding premature conclusions using different approach
The estimation of the informal economy for the
Republic of Macedonia up to 2008 has been
reported consecutively as seen in[ [13] ],[ [14] ] or in a
more general view as in.[ [15] ] In our work we
considered the evaluation of this parameter for a
narrow period of time defined as [2004, 2016].
Some specifics of the calculation for this period
were reported in.[ [8] ] Herein, we are interested on
the size of informal economy and the effect of
some factors on it. We have analyzed many of
them, but purposely we will discuss the corruption
effect on the informal economy. In this approach
we considered the calculation of the informal
economy using discrepancies models and used
those estimations to check the possible
relationship between the latent variable and
factor-like ones. The most highlighted finding we
observed for some categorical variables which we
will discuss shortly below. When applying
unemployment model of informal economy using
yearly data we obtained that the result is quite
similar with the estimation using GDP
discrepancies as seen in the Figure 1.a. It has the
same trend as the corruption index as seen in the
Figure 1.b. Therefore in mechanical view, this
variable is expected to influence our latent
variable. In practice, there is a common belief that
correlation is indicator of causative.
fig-24ce7dc981bfc31da8aa60b62b2cc348
a. Percent of Informal GDP. Orange line,
according to GDP discrepancies, blue line,
according to Unemployment model
fig-edb7f3041422f45bd7c1a1576ed28539
b. Relativized Corruption index (orange line )
and Informal Economy (black line)
The same result is obtained in the case of the
variable measuring administrative performance of
the government as the ratio Budget Deficit/GDP.
In our analysis we commented those findings as
indicators for possible relationship but more proof
is needed to conclude. Next we use the linear
modeling of the type (3) extended as discussed in
[ [16] ] and other models of type (4) using forms
suggested on the literature. We obtained that the
regression and the statistics for the variable “index
of corruption” didn’t confirm the direct
relationship with informal economy and therefore
the high positive correlation seen in the Figure 1 is
considered as false in our case. In general this is
not novel as seen in [ [17] ]. This suggested that the
discrepancies method is not accurate and should
be revisited. The other possible reasons could be
the nature of categorical variable. Similar results
have been obtained for the other parameter
analyzed the quality (performance) economical
governance. Adding to the remarks above, in this
case we proposed to use more tinny series based
on quarter or even monthly data. But from our
actual system perspective, we supposed that linear
models would be applied if there is no regime
change in the period examined, so it would have
been checked with priority.
2.2. Identification of the best interval where a linear model is reasonably applicable
We analyzed the set of variables proposed in [ [1] ]
in the framework of CDA or MIMIC model. The
trend of variables (logarithmic variables) showed
two special points, one was located around 1990
and the other one in the zone [2002, 2004]. We
observed that the behavior of the informal
economy calculated using Cagan model as given
in [ [1] ] etc., clearly differs to the one variable that is
expected to be co-linear, as the rate of
unemployment or example, Figure 2.a. Using the
reviewed CDA model proposed by Tanzi, we
obtained the same behavior which support the
evidence of a strong change in-betweens the years
2002-2004. Therefore, this consists in a special
point that should have been avoided form the
series used in linear forms as (3). Moreover, the
estimation with CDA in the full interval [1998,
2014] showed an undervaluation of informal
economy, because the level of 20% is
characteristic for the developed country as
reported in [ [7] ] which is not our case. We thought
that the major cause for the inconsistencies of the
value of informal economy estimated for interval
[1998, 2014] as evidenced in Figure 2 was related
to apparently regime change in 2002. So, a better
tactic for linear regression would have been the
use of data series that do not included records
from this time zone.
fig-a0cd9b45dacbf18b7664c1212deab025
a. Informal Economy by Unemployment method
fig-6f3fc78b6e163b2ba2b6ff8c14264729
b. Informal economy by CDA in Tanzi version
By such a correction of the reference interval we
have improved the calculated the informal
economy in the period [2004, 2014] as given in
Figure 2.b. In this case we obtained a high value
of it around 38% of GDP near 2010-2011 and a
decreasing trend form this date as seen in the
Figure 2.b. the average level of informal economy
has been found higher than 20% and this result
has been supported by other models too.
2.3. Assessment of typical variables representing particular categories on models
Another issue has been initiated from unexpected
findings for the effects of some factors. So, in
calculation of the type (3) coefficient of linear
regressions for some categories of Taxes were
found negative that seems to be a meaningless
result. In principle this is possible as discussed in
the reference [ [2] ], [ [3] ] and others because
economical systems are complicated in the best
approach, and sometimes behave as complex in
the full sense. A corollary of this property is that
there is no rigid model for all cases, as is proven
to be true in many applications. In such a case we
operated according tactics mentioned above to
exclude “exterior troubles”. Next we extended the
idea of an adaptive application of the models
recommended in the literature as [ [6] ] or [ [5] ] by
considering a set of variables for each category
involved in equations of the type (3) or (4). Thus,
we searched empirically to find which concrete
variable could be exactly in the role described by
the appropriate category in the equation. So, in
CDA model the variable “Taxes” appears in
relation as ln(1+Tax.Rate) [ [1] ]; variables “Interests
Rates” appears directly in equation of the type (3),
other parameters appear in logarithm Figures etc.
Hereto we realized a generalized regression
procedure of a linear approach for ln(C/M)
including all categories of variables and all
variables under examination. The series that
“survive” statistical test for regressions including
a satisfactory ratio of error to the average value
evaluated as seen in Table 1, have been qualified
as the appropriate term for final calculation within
the model of the type (3) or (4).
fig-23371df02654dab8175dda150e244b3d
In the Table 1 we observed that all variables in the
regressions have good pvalues, but VAT has a
good error to value ratio and good statistical
fitting parameters. The variable “Average Taxes”
that appears in standard models of the type (3) and
its Tanzi equivalents seems to be not the best
choices. In the other side, the VAT has good
statistics (pvalue is obtained lower than standard
thresholds) and good error to vale ratio. Moreover,
we obtained that in some regressions of type (3)
the coefficient of the term ln(1+AverageTaxRate)
has negative value that is theoretically wrong. The
two other terms including VAT or Total Taxes
have had positive sign in the regression so each
one could play the role of “Taxes” in theoretical
model. Taxes are expected to simulate the
informal economy so the sign should be positive.
In the same way we have to decide which element
of the category “Interest Rate” plays the role of
the Interest rate for Deposit as an expected
inhibitor effect in informality in original model
and so on.
2.4. Remarks for the variable of type currency
The last element we addressed herein was the
Indicator variable. In theory the indicator of
informal economy is money aggregate. The
question herein is to identify which one of them
represents at the best the informality for the
economy. Again, the direct use of series has given
acceptable values but not consistent with some
expectation. The most important indicator for
informal economy is the currency out of deposits
[ [1] ] so we discussed it more thoroughly. We
considered the fact that the amount of money
entering in cash the country in the form of
remittances is considerable,[ [18] ],[ [19] ] In general,
those inputs are found usually in Euro currency
which has been used as payment tool as well,
behaving temporally as national currency itself
(Denar) and therefore we proposed to use it as part
of total money in circulation by replacing C/C+R
in the CDA model. Notice that by direct use of
models we obtained a lower than expected level of
informal economy whereas if including
remittances in C in has shifted up by a few
percent. But the most intriguing part of these
discussions comes from finding represented in the
Figure 3.a where it is seen that C/D has a
decreasing trend in all the period [2001, 2015]
whereas it is supposed to be proportional to the
informal economy.
fig-e848d969122881065f0d3eb4f770f579
By red line, the ratio of Currency
(C)+Remittance to Deposits (D); blue line,
C/D ratio
fig-23de54f5062f17e3af75686ce7225b09
b. In orange, informal economy by simple Cagan
model. Blue line, Currency +Remittances
This last was not reported to show constant
decreasing trend in this period and therefore the
initially assumed indicator variable C/D would
better be (C+R)/D. Hereby we noticed the
improvement of the calculation but again the
change from year to year is high, in some case
around 5% that is not theoretically supported. This
last remark suggests again the use of tiniest series
which we reported in.[ [8] ]
3. Calculation of informal economy for the
Republic of Macedonia
We have considered all precaution steps analyzed
above in the final estimation of the informal
economy of the Republic of Macedonia. The
opportune period for calculation in [1996, 2016] is
considered the narrower interval [2004, 2014].
Other findings and comments considered have
been reported in [ [8] ] as typical for the application
of linear models, CDA, SCR or even MIMIC in
our concrete system. Next the variables being
qualified as more representative of a category
related to the linear form (3) have been used in the
calculation based on simple currency ration (SCR)
and Currency Demand Approach. The result
showed significant improvement on the
determination of major causes for informal
economy in the country. Consequently we were
able to better selection of the set of variables to be
used as factors and indicators in the more
advanced MIMIC model. We observed that those
steps improved the calculation.
fig-ebc16ebe259cb2d0f7ccc6fc25d44937
a. Informal Economy by MIMIC 8-1-3 model
:
fig-3882c503233b9e9c62b1665fc6ea3c6d
b. Reproduction of the indicators: yellow line,
unemployment rate, blue line, ln(GDP); red line
logarithm of narrow money,
Now the latent variable showed slighter changes
say around 1%-3% from year to year and
moreover the reproduction of the indicators based
in model’s parameters is obtained qualitatively
good (quite correct in logarithmic scale) as seen in
the Figure 4.b. Next, the full analysis of the
system has been performed straightforwardly.
Thus, variables as Remittances, GDP per capita,
VAT, GDP deflator, Taxes over Personal
Incomes, and Interest Rates of Deposits have been
identified as key factor in the size of the informal
economy for the country. Indicators of the
informal economy were identified to be the GDP
per Capita, the Narrow money M1 and the Rate of
Unemployment. The unregistered economy has
reached 35% of the GDP near the year 2010 and
now is around 31-33% of the GDP. Oscillations
continue to be present on the trend of the annuals
value estimated, so the analysis needs for more
improvement which remains to be addressed in
our future works.
4. Conclusions
We realized to find the most appropriate time
interval to implement linear modeling in the
calculation of the informal economy for Republic
of the Macedonia. We obtained that the set of
variables that mostly influenced the informality
and shadow economy could be assigned by a step
by step selecting procedure. In particular, we
observed that the corruption and economic
performance of the government have a
complicated effect, whereas Remittances, Value
Added Taxes, Taxes in the Incomes, Interest of
Deposits and GDP acts as a set of factors that are
responsible for the size of informal economy for
the period [2004,20014] . We identified GDP per
Capita, the Narrow Money and the
Unemployment as key indicator of the informal
economy for the period exanimated. The
underlined trend of informal economy for the
EM-2018-673
country in the period [2004, 2014] showed the
peaks at 35%-38% around the year 2010 and now
it is decreasing toward a stationary state to the
around 33% of the gross national production. We
obtained herein that the majority of economic and
political measures undertaken recently in the
framework of informality reduction have worked
in the aimed direction. From dynamical point of
view we obtained that the overall contribution of
some variables in the informal economy is
complex and therefore their quantitative weight
would probably change in the futur
References201310.1017/cbo9781139542289SchneiderFriedrichEnsteDominikHThe Shadow Economy111122201110.1007/978-88-470-2062-7_6PescatoriMarioPrevenzione e trattamento delle complicanze in chirurgia proctologicaTumori del retto e dell’ano201210.2139/ssrn.2057864ArdizziGuerinoPetragliaCarmeloPiacenzaMassimilianoTuratiGilbertoSSRN Electronic JournalMeasuring the Underground Economy with the Currency Demand Approach: A Reinterpretation of the Methodology, with an Application to Italy253725552008-oct10.1080/00036840600970195Dell’AnnoRobertoSolomonOffiongHelenApplied EconomicsShadow economy and unemployment rate in USA: is there a structural relationship? An empirical analysis200310.5089/9781451842432.001andEbrimaFaalIMF Working PapersCurrency Demand, the Underground Economy, and Tax Evasion: The Case of Guyana10.4337/9780857930880.00007SchneiderFriedrichBuehnAndreasMontenegroClaudioEShadow Economies All Over the World: New Estimates for 162 Countries from 1999 to 2007201010.1596/1813-9450-5356MontenegroClaudioESchneiderFriedrichBuehnAndreasShadow Economies All Over The World : New Estimates For 162 Countries From 1999 To 20072018-jul10.18535/ijsrm/v6i7.em12AsaniDashamirPrengaDodeInternational Journal of Scientific Research and ManagementA practical strategy to improve econometric modeling–a case study for informal economy on the Republic of Macedonia91962013-feb10.25103/jestr.061.18FangYinglanHanandXianfengHanBingandJournal of Engineering Science and Technology ReviewResearch and Implementation of Collision Detection Based on Modbus Protocol10.4337/9781847205858.00017SriskandarajahDhananjayanMigration and Development: Managing Mutual Effects10.4337/9781784717995.00015Tackling the shadow economy and shadow labour force3556199810.1007/978-3-642-72032-1_4MerzThomasWeb Publishing with Acrobat/PDFCreating PDF Files201610.18411/d-2016-154http://ljournal.ru/wp-content/uploads/2016/08/d-2016-154.pdf201110.1596/1813-9450-5770CeleskaFrosinaGligorovaViktorijaKrstevskaAnetaMacroprudential Regulation of Credit Booms and Busts: The Experience of the National Bank of the Republic of Macedonia6222015-dec10.1515/seeur-2015-0025OsmaniRufiSEEU ReviewThe level of the shadow economy, tax evasion and corruption: The empirical evidence for SEE countries256271201710.4236/me.2017.82018OuédraogoIdrissaMModern EconomyGovernance, Corruption, and the Informal Economy2152382009-sep10.1007/s11127-009-9513-0DreherAxelSchneiderFriedrichPublic ChoiceCorruption and the shadow economy: an empirical analysis1983-jan10.1515/bd.1983.17.8.679aBibliotheksdienstOsteuropa-Bibliothekare tagten in Regensburg1983-aug10.2307/2061244SellRalphRDemographyAnalyzing Migration Decisions: The First Step–Whose Decisions?5262017-nov10.24193/rvm.2017.10.07AndonovaVesnaGarvanlievaNikolovandMarjanSelmaniDenizMitevskiIgorKundovskaFatimaOsmanovskaandVirgil Madgearu Review of Economic Studies and ResearchThe Policies and Measures for Self-Employment and Entrepreneurship in Macedonia among the Roma Community