My findings suggest that more than 90% of the decline in informal competition can be attributed to socio-economic changes. Focusing on the perceived heterogeneity, the main contributors to the decline in the pace of informal competition are an educated workforce, access to finance, losses due to theft, transport as an obstacle to current operations and the service sector. Recovery of the economy and commodity prices contributed to the decline in informal competition rate in Kazakhstan since 2009.

According to La Porta and Shleifer (2008) there are three main views on the role of informal firms. The focus of this paper is to assess the determinants of informal competition in formal enterprises. Using BEEPs data, I analyze the determinants of the decline of informal competition in Kazakhstan over time due to apparent heterogeneity and structural change.

The size of the informal sector in any country is a good indicator of the level of competition that formal firms face (Gonzalez and Lamanna, 2007). Schneider and Williams (2013) discuss that the size of the shadow economy is negatively related to economic cycles. The size of the shadow economy in the 1990s given in Table 1 was obtained using the electricity consumption method.

According to the comprehensive plan to counter the shadow economy, one of the main causes of the informal economy in Kazakhstan is the migrant labor force of the Commonwealth of Independent States (CIS).

## Fixed Structure Analysis

I use fixed structure and changing structure analyzes to assess where the main driving force behind the decline in informal competition in Kazakhstan is due to structural change or observed heterogeneity, and to explain to what extent the decline in informal competition can be explained by changes in the levels of underlying factors: corruption, taxes, crime, access to finance and the level of education of employees and socio-economic changes. In words, I estimate the sum of the change in informal competition due to changes in the independent variables by multiplying any change in the probability of informal competition due to the change in variableXi by the actual average change in that variable occurring between the two periods.

## Changing Structure Analysis

Standard principal component analysis (PCA) is based on the Pearson correlation matrix and assumes that the variables are continuous and follow a multivariate normal distribution. I use PCA to understand which variables contain the most variance, since my model includes variables that are dichotomous, ordinal, and continuous, PCA is performed using a polychoric/polyserial correlation matrix. According to Kolenikov and Angeles (2009), in the calculation of the polyserial correlation, the probability for the latent variable x1 with the basic standard normalox∗1 is discretized according to the thresholds α1,0 = ∞ < α1,1 <.

Once the correlations are estimated, the next step is to continue the PCA by solving its own problem for the estimated correlation matrix. In Stata, the polychoric correlation matrix considers the types of variables, so when the variables are binary, the polychoric correlation is calculated; when variables are continuous and categorical, polyserial correlation is calculated, and if all variables are continuous, Pearson correlation is considered. The KMO score can be interpreted as poor, meaning that the variables share a low level of common factors.

However, since my data fail to reject the Bartlett's test of sphericity with null hypothesis that the variables are correlated with each other at a 1% level of significance, there is evidence in favor of performing polychoric PCA. However, all the factors contribute to the total variance, and according to the scree rule, I include all the variables in my regressions. Since the variables of interest are not highly intercorrelated, given low KMO, along with the results of correlation matrix, I proceed with all 11 variables for my fixed and changing structure analysis.

The first components relate to access to finance, labor regulations, losses due to theft and transport as an obstacle to firms' operations.

## Fixed Structure Analysis

The estimates show that holding everything in their overall sample means that labor regulations as well as firm size and southern region contributed to an increase in the rate of informal competition by 5.2%. Holding constant firm size, labor regulations, and regions variables, the level of university-educated labor force, access to finance, theft, transportation as a major barrier, and the service sector contribute to the 5.4% decline in informal competition. Although some factors have caused the degree of informal competition to decline and others have further increased it, I am unable to explain most of the current decline with these variable changes.

Furthermore, I estimate whether some of the relationships between informal competition and its determinants have changed over time and whether this might help explain the decline in informal competition.

## Structural Change Analysis

Firm size, location and educated workforce contributed to the decline of informal competition; the southern region and the service sector contributed to its increase. Because several coefficients differ in their effect on the level of informal competition, I further assess the relative importance of these structural changes versus changes in variables in explaining the decline in informal competition between periods by seeing how well the 2008–2009 and 2012 coefficient– Explanatory variable values for 2013, they predict the level of informal competition for the period 2012-2013. The difference between the predicted level of informal competition in 2008–2009 and the actual level in 2012–2013 is the proportion of the decrease in informal competition attributable to changes in the variables.

This difference between the predicted 2012-2013 informal competition rate and the actual 2012-2013 rate is the portion of informality decline due to structural change (i.e. coefficients). Thus, I am interested in how well I would have predicted the informal competition rate given just the knowledge. In the actual rate of the informal competition was 35.1% and the predicted value was estimated at the level of 38.9%, while the actual rate of informal competition decreased to 32.9%.

I conclude that a significant part of the decrease in the level of informal competition between 2012-2013 is the result of changes in the structure of informal competition. Thus, structural changes between 2008 and 2012 and the determinants of informal competition in 2012 contributed to the reduction of informality. The loss of government revenue in the form of income tax and social tax due to informal employment amounted to USD 70 million in 2014.

The level of informal competition remains in retail (bazaars), agricultural and construction sectors due to seasonal work and the large number of migrants in these seasonal jobs (World Bank, 2014). This study identified several factors that contributed to the decline in informal competition between 2008–2009 and 2012–2013 in Kazakhstan. The research also assessed the factors that prevented the level of informal competition from decreasing further.

The decrease in average firm size and increased entrepreneurship in the southern part of Kazakhstan prevented the reduction of informal competition from being even greater. An educated workforce, access to finance, losses due to theft, transport as a major obstacle and the service sector contributed the most to the reduction of informal competition. As the BEEP survey did not cover all industries and sectors, it is possible that the level of informal competition is affected by factors other than those reviewed here.

To curb the informal economy and the speed of informal competition, it is worth working on improving the institutional and legislative framework in Kazakhstan. Yasser Abdih and Leandro Medina Measuring the Informal Economy in the Caucasus and Central Asia".