The effect of economic sanctions on corporate cash holdings and profitability : the case of Russia

(1)

Master Thesis

The Effect of Economic Sanctions on Corporate Cash

Holdings and Profitability: The Case of Russia

Amsterdam Business School

Nikita Sergeechev

10630503

MSc Finance, Quantitative Finance track

Supervisor: Dr. S.R. (Stefan) Arping 

(2)

Statement of Originality

This document is written by Nikita Sergeechev and I declare to take all the responsibilities concerning the document.

I state that the work and the text presented in this document is original and no sources, other than mentioned were being used. The Faculty of Economics and Business, namely Amsterdam Business School is only responsible for the supervision of the work and not for the content.

(3)

Abstract

Purpose:

The purpose of this paper is to investigate the effect of Economic Sanctions on corporate cash holdings and profitability of the various large companies in the Russian Federation. In this research, several parameters in order to answer the research question will be introduced. The primary extent of the effect of the economic sanctions is conducted with the help of two measures: the ratio of the cash holdings to the total assets and the secondary on the return of assets.

Design/Methodology/Approach:

The Panel data Difference in Difference (DID) regression is being used to investigate the relationship between the binary variables for the economic sanctions and the control variables, which might have an effect on dependent variables, for instance, the cash ratio of the Russian companies and their return on assets. The time and firm fixed effects are being used to answer the question in the research. These methods are supplemented by the random effects panel model and descriptive analysis of the data, as well as some mean equality tests for sub-samples.

The data for this study is quite recent to the moment when this research is finalized and covers the quarterly periods from the year of 2010 to 2017.

Findings: a positive effect of the economic sanctions on the firm’s cash holdings and an adverse

impact of the economic sanctions on the firms expected profitability.

Keywords:

Russia, Economic sanctions, cash holdings, profitability, fixed effects model

Paper type:

(4)

Abstract……….………...3

1. Introduction………...5

2. Literature review……….6

3. Which and whose economic sanctions are studied in the paper….14

4. Hypotheses……….…..…15

5. Methodology……….…..15

6. Data………..….20

7. Results………..………...23

8. Empirical Findings……….…..…38

9. Conclusion……….……….……….38

10. Discussion……….………..…...39

11. References….………...….40

Appendix………..42

(5)

1. Introduction

During the last decade of the twentieth century, Russian Federation has experienced problems with a transition from the centrally planned economy to the market economy. However, at the beginning of the 2000s Russian economy has managed to overcome such difficulties and continued to grow till the second decade of 2014. Even the financial crisis of 2008 has not changed this growing trend. The main reasons and fuel for the growth were the increased prices of oil and gas, which have a significant share in GDP. The situation has seemed very fortunate for the country until the heightened tensions with the Western countries has started to grow.

As it has already been mentioned, in 2014 the geopolitical circumstances for Russian

Federation have turned the corporate sector into an economically depressive phase. The Western financial and economic sanctions have not only brought severe difficulties concerning trading opportunities for various Russian companies (Golikova, Kuznetsov, p.49) but have also affected the financial and mining sectors in Russia, namely state-controlled banks, oil and gas companies. They suffered from the reduction of cash flows and deteriorating funding conditions (Gurvich & Prilepskiy, 2015).

The sanctions, which have been implemented by such states, as the European Union, the United States, Canada, Norway, Iceland, Canada, Australia, and Japan were not only introduced for the specific companies in the Russian Federation but also were set up against specific

politicians, people in business or other notable people. However, the goal of the paper will be to estimate the effect of the sanctions on the Russian corporate sector. To be more precise, this effect will be examined through the estimation of the cash holdings of the Russian companies and their profitability.

The analysis done in this study has a value nowadays. The following can be said since previously the period to make a similar study was relatively short, and it has not been possible to correctly estimate the actual effect of the sanctions on the various financial parameters.

However, by the beginning of 2018, it has been almost four years since the first sanctions on the Russian corporate sector have been introduced, and there is more evidence for the analysis. Moreover, it was still not apparent what the long-run impact of the sanctions will be. Apart from that, the uniqueness of the research is that it does not only consider the publicly traded

companies, but also the private ones, especially in the military and defense sector. Last, it entirely focuses on the corporate journal entries and carefully investigates the effect of financial restrictions on the financial parameters of companies. Therefore, at this point, the effect on financial parameters may not only be seen but also be carefully investigated to a full extent.

(6)

2. Literature review

The literature review based on the most relevant research papers on similar topics will be constructed in a specific way. First of all, a short description of the Russian economic and financial situation will be provided. Later on, the definition of the word ‘sanctions’ and its explanation in the context of the Russian corporate market will be introduced. Furthermore, a brief description of what economic and financial sanctions are and how they might affect the Russian corporate market will be given. Finally, the existing models of the corporate parameters will be discussed. These models form a methodological basis and lay a foundation of the approach used in the paper.

2.1

The situation in the Russian economy and corporate market (1991-2014).

As it has already been mentioned in the introduction, after the collapse of the USSR in 1991 the Russian economy over these past 27 years (since the state was formed in 1991) has not been in an extremely stable phase. It was moving from the transitionary economy (the economy during the 1990s) to the full market economy, as it seems to be now. During the beginning of that process, the Russian capital market has obtained several features of an emerging market (Mirkin, Y., Kuznetsova, O. & Kuznetsov, A., 2013). More elaborately, these features included a limited range of instruments, low liquidity, and operational volumes, dependence on international investors, high volatility. At that stage, Russia was not able to cope with the new characteristics of the financial markets and speculative attacks. As a result, in 1998 the default on the short-term bonds took place. Surprisingly, the recovery phase of the Russian economy after the turndowns has happened fast, and the main reason for such change was an enormous increase in the oil prices (Mirkin, Y., Kuznetsova, O. & Kuznetsov, A., 2013). The Russian economy kept on growing by great amounts since 1999 till 2014 and not even the financial crisis of 2008 has significantly influenced such tendency. However, as Mirkin, Kuznetsova, and Kuznetsov point out, the Russian financial market kept on remaining highly volatile with a significant tendency on speculative activities with a high emphasis on the operations, influence, and participation of foreign residents. It was also a fact that the most crucial Russian companies were financed from abroad and the Western banks served a vast majority of them. As it comes to talk about the Russian trade balance, it may be said that it has become more independent on the imported highly technological imported products (Shakina, E., Barajas, A., & Molodchik, M. (2017). Therefore, most of the Russian products were lacking quality. The assets of the Russian companies were depreciating, and this could also be explained due to the insufficient levels of human resources and marketing. Eventually, the country could only keep its trade balance positive, due to the increasing oil prices and increased oil extraction.

(7)

Figure 1. Russian Ruble to US dollar daily exchange rate (March 1996- March 2018)

Source: Thomson Reuters

Therefore, it seemed evident that if the Western presence was reduced and if the oil prices went down it would significantly harm the Russian financial market and its economy.

Figure 2. Russian yearly GDP in PPP in current prices (1995-2017)

Source: Federal State Statistics Service, Russia/Thomson Reuters

2.2

The current situation in the Russian economy and corporate market (2014-2018).

The year of 2014 has brought very disbalancing political circumstances to Russia, which have resulted into Economic instabilities, namely the decrease of oil prices, depreciation of the Russian

0 10 20 30 40 50 60 70 80 90 3/ 6/ 96 3/ 6/ 97 3/ 6/ 98 3/ 6/ 99 3/ 6/ 00 3/ 6/ 01 3/ 6/ 02 3/ 6/ 03 3/ 6/ 04 3/ 6/ 05 3/ 6/ 06 3/ 6/ 07 3/ 6/ 08 3/ 6/ 09 3/ 6/ 10 3/ 6/ 11 3/ 6/ 12 3/ 6/ 13 3/ 6/ 14 3/ 6/ 15 3/ 6/ 16 3/ 6/ 17 3/ 6/ 18 0 500 1000 1500 2000 2500 3000 3500 4000 4500 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

(8)

ruble and the introduction of the economic sanctions by numerous countries. Such countries have been from all over the world and include the United States, countries from the European Union, Canada, Australia, Japan and several others. According to Oxenstierna and Olsson (2015), the reason for such sanctions was the fact that the Western countries claimed that the Crimean Referendum after which the state has joined Russia was illegitimate, and Russia has also not kept Ukrainian sovereignty by supporting the pro-Russian separatists in Eastern Ukraine.

Furthermore, comparing the current Russian Economic and Financial trends to the ones of 1991-2014 it can also be inferred that no reforms have taken places in the Russian economy and corporate sector. All that economic downturns could be the reasons of stagnation of the Russian economy. Such trend remains to be the case until the beginning of 2018.

Figure 3: BRENT crude oil prices (1991-2018)

Source: Thomson Reuters

2.3

What are the economic and financial sanctions imposed by different countries?

According to Verbeel and Markus (2015), sanctions are a tool of coercion and are aiming to change the actions of a target country and to prevent it from doing undesirable activities. Moreover, the change in the behavior of such country will be taking place due to bearing additional direct material costs and, such sanctions can be a reason for such country to change the behavior.

The European Union has been very active in implementing financial and trading sanctions

0 20 40 60 80 100 120 140 160

(9)

against the Russian Federation. For instance, the law, which came to power on July 2015 has forbidden EU nationals and companies to buy or sell new bonds, equity from several biggest Russian banks (Verbeel & Markus, 2015), which include Sberbank, VTB, Gazprombank, Rosselhozbank and VEB and three major Russian defense companies, for instance “Kalashnikov Concern”. Moreover, the European banks could no longer provide loans to five important Russian state-owned banks and the issuance of financial instruments have been prohibited as well. Furthermore, an embargo on trading the military arms has been put into force. Last but not least, the trades of technological and energy equipment with Russia has become a subject to prior assessment by the members of the Union. As a result, major Russian oil companies have faced problems, as they could no longer receive technological items for the oil extraction projects in the Arctics and the shale (Verbeel & Markus, 2015).

According to Orlova (2016), the sanctions that have been imposed on the Russian banks and companies can be divided into three groups. The first group can be defined as “sectoral sanctions”, as the U.S Sectoral Sanctions Identifications have also defined it. Such group includes specific firms that have been denied in access to the American and European credit market but can still process payments of their clients and their own payments. The second group is the ‘Specially Designated Nationals List’. Such group includes several private banks and a small number of companies from different industries, mainly from the defense sector. Such companies face the prohibitions in foreign exchange payments and one only option for them would be to change their business orientations. The last group is dealt with the changes in the business environment, namely the changes in attitudes towards the Russian capital and the way in which the Russian issuers are taken care of. As a result, their financial regulation procedures have become more severe.

2.4

The opinion of the scientists regarding the Economic sanctions on the Russian Federation.

There have not been any similar researches in the light of financial sanctions imposed by the Western European countries, like Canada, the USA, Australia or New Zeeland. Therefore, the econometric analysis on the selected topic will also be unique. Although, the scientists still have a strong opinion about how the sanctions had an effect on the Russian financial market and economy and it will be described below.

According to Christie (2015), the design of the sanctions was not intended to harm the Russian economy and the Financial sector catastrophically, but more to have a long-term impact unless the Russian political would change. Such sanctions should harm more the Russian party, rather than

(10)

the western countries and furthermore, they should be balanced across sectors and member states. Moreover, they should also be easily reversible and scalable and last, but not least their defensibility and ease of enforcement should be at a high level.

Oxenstierna and Olsson (2015) state that the companies who faced the most severe restrictions on trade were the ones dealing with the financial transactions in such sectors as finance, energy, and defense. Moreover, such authors mention the fact that sanctions were causing numerous troubles since the sanctions have brought significant difficulties for the state banks and large companies, which need to refinance their debt. Last, but not the least the authors have also discussed the issue with trading opportunities for the companies from the Russian Federation and have come to a conclusion that the reorientation of trade has been taking place (p.46). For instance, Russian oil and nuclear companies are currently in a transitionary phase from the EU (European Union) market, which has implemented such sanctions to the Asian ones, for example, China or Turkey. The same can be said about military-oriented firms, who were affected in a way they could no longer cooperate with Western Defense system and dual-use technologies. As a result, it will take time for the profits of such companies to be back to the pre-sanctions period.

Orlova (2016) talks about the significant effects for the Russian financial market, namely banks. She claims that several entities would need to restructure their balance sheets and switch to the internally financed sources or to find additional sources overseas, for instance in China and Korea. She has also pointed out that financial regulation procedures for the Russian companies, especially the Russian banks have become more severe. As a result, a notable problem with the transaction processing (compliance) has been taking place. Moreover, the Russian issuers of debt securities have also dealt with the changes in risk management protocols of Western banks, and this has resulted into a business environment deterioration for the Russian firms. Furthermore, many foreign banks have decided not to increase the presence in the Russian capital markets. Additionally, the sanctions had a negative impact on the potential investors from overseas, due to an increased concern about the credit portfolios of the Russian banks. Lastly, the domestic currency market in Russia has suffered a lot. Russian companies faced severe problems in paying back their debt in foreign currencies, and this has resulted in a dollar liquidity deficit. She has concluded that there is a growing tendency of the government banks expansion, as there is a relative risk minimization in such entities.

Christie (2016) have concluded that the outcome of the financial sanctions would be that Russian entities, namely banks will tend to reimburse all the external debt progressively. Therefore, the vulnerability for such firms will tend to decrease. Moreover, the banking sector

(11)

will be facing underinvestment, so the competitiveness and the productivity of such companies will decline bringing some liquidity troubles to the Russian corporate liquidity market. Finally, an overall negative effect on the Russian economy would also be taking place.

On the other hand, Gurvich and Prilepskiy (2015) have come up with the idea that the sanctions had a severely adverse effect on net capital outflow and foreign direct investment. Apart from that, Russian firms who were under the threat of sanctions faced limited borrowing opportunities. Therefore, it had a significant adverse effect on the amount of cash in such companies. In fact, this is mostly the case for banks and companies in the energy and military-industrial sectors, which are publicly held.

As for the paper of Golikova and Kuznetsov (2017), the authors have come up to the idea that the impact of the sanctions on the Russian manufacturing sector will become more visible within a more extended period, namely medium to long-term.

Concluding the overall review of the articles mentioned above, it can be said that such authors have described the consequences for the Russian financial market overall but have not gone through the econometric analysis. They have only covered the basic descriptive statistics not only of the financial sector but the economic sector overall. The following results can be explained that this topic is relatively new and only now the moment comes when the set of data can be analyzed in further details.

Several authors have also come up with an econometric analysis regarding such sanctions, for instance, Pak and Kretzchmar (2016). They have analyzed the impact of state ownership and specific business variables on bank capitalization during the time of escalating sanctions using the Panel Study regression. They have come up to a conclusion that the sanctions would adversely affect liquidity, mainly if the poor credit management will persist.

Hoffmann and Neuenkirch (2017) have analyzed the impact of the Russian-Ukrainian conflict on the Russian and Ukrainian stock returns, but the time span has been relatively short and included the period from November 2013 to September 2014. They have deduced that the Russian financial system has indeed suffered from such a conflict. It was seen that the Russian stock market returns decreased by 21 basis points.

2.5

The determinants of corporate cash holdings or the profitability parameters by other authors

There have been several authors, who used to discuss the determinants corporate markets, namely the corporate cash holdings or the profitability parameters.

(12)

For instance, Opler, Pinkowitz, Stulz, and Williamson (1997) have used the American publicly traded firms from 1971 to 1994. They analyzed the determinants and implications of corporate cash holdings with the help of panel regression. The authors have used such variables, as investment opportunities, regulation, firm size, financial leverage, dividend payout, cash flow and several others. By the end of the research, they were able to infer that firms with substantial growth opportunities and firms with riskier activities tend to have more cash than other firms. They have also evaluated for firms to invest easily they prefer to hold more liquid assets if the cash flow to investment is low and when the funds from outside are too costly.

Another author, namely Al-Najar (2013) has used a slightly different approach to estimate the financial determinants of corporate cash holdings in the emerging markets. He used several ratios, which include return on equity, leverage ratio, liquidity ratio, dividend payout ratio and a binary variable for the size of the firm to determine the effect on the cash ratio. The author manages to find the relationship between such independent variables and the cash ratio parameter. Therefore, he concludes these variables can indeed be the determinants of cash holdings. The core of such panel regression will be later used in this research.

As for the profitability analysis, the paper of Goddard, Tavakoli, and Wilson (2005) seems attractive since the scope of the companies taken into account is relatively big and touches various samples in the manufacturing and service industries. The authors have used a dynamic panel to investigate the relationship between the non-current liabilities plus loans divided by shareholder funds (GEAR), market share, the ratio of current assets to current liabilities and the lagged values of return on assets (PROF at time t-1 and t-2) on the current return on assets (PROF at time t), which is measured as interest plus net profit before tax divided by total assets on the natural logarithm of total assets (ASSET). Besides, the country and firm effects have been introduced. The analysis has been performed in several European countries, which include Spain, United Kingdom, Italy, France, and Belgium. By the end of the paper, the authors were able to obtain several interesting findings, which include a negative relationship between the GEAR and profits, a positive relationship between the liquidity ratio and profitability, negative size profitability relationship. Last but not least the relationship between the market share and the profitability is also positive. Several ideas from this paper will be used to construct the second regression in this work. Such regression in the paper will determine the effect of the economic sanctions on profitability.

(13)

Figure 4. The average Return on Equity (Net Income over Total Equity) in the Russian corporate market measured in percentages. The period studied on the graph is from 2009 to 2017.

Source: IMF - Financial Soundness Indicators

Figure 5. The average amount of return on assets (Net Income to Total Assets) in the Russian corporate market measured in percentages. The period studied on the graph is from 2009 to 2017.

4.06 12.49 17.32 17.89 13.98 7.53 1.96 9.84 7.94 0 2 4 6 8 10 12 14 16 18 20 2009 2010 2011 2012 2013 2014 2015 2016 2017 0.72 2.04 2.47 _2.39 1.87 0.95 0.23 1.2 1.01 0 0.5 1 1.5 2 2.5 3 2009 2010 2011 2012 2013 2014 2015 2016 2017

(14)

Figure 6. The average amount of 'Liquid assets to Total Assets’ in the Russian corporate market measured in percentage. The period studied on the graph is from 2009 to 2017.

3. Which and whose economic sanctions are studied in this paper?

The sanctioned companies, which will be discussed in the following research will be based on the sanctions imposed by the authorities of the United States. In more details, such penalties were the ones implemented by the US Office of Foreign Assets Control (OFAC) and was the specified Executive Order 13662. The following financial penalties were imposed for several Russian firms and entities, due to the Russian aggressive and vulnerable actions against the territorial integrity of Ukraine. Such penalties imply severe problems for such Russian

companies when dealing with American laws and companies. Such orders have been published in March 2014 and have become active to the list of the specified companies in the July of 2014.

Based on the OFAC Executive Order 13662 and the four directives the US entities could no longer deal with transactions in, provision of financing for, and other dealings in new Russian debt of longer than 30 days maturity or new equity of the financial sector companies. The same debt restrictions but now of longer than 90 days were applied to energy, materiel companies. Lastly, this Order has also implied the prohibitions of the US parties to participate in “the provision, exportation, or re-exportation, directly or indirectly, of goods, services (except for financial services), or technology in support of exploration or production for deep water, Arctic offshore, or shale projects that have the potential to produce oil in the Russian Federation, or in maritime area claimed by the Russian Federation and extending from its territory, and that

30.37 28.99 25.51 24.65 21.67 23.35 26.53 23.58 25.36 0 5 10 15 20 25 30 35 2009 2010 2011 2012 2013 2014 2015 2016 2017

(15)

involve any person determined to be subject to this Directive, its property, or its interests in property.” More information about the data selection for this paper will be provided in section 6.

4. Hypotheses

The primary hypothesis for the following topic will be related to the effect of the Economic sanctions, which were introduced in 2014 on the cash holdings, namely the cash ratio. The following hypotheses will be based on the empirical tests performed in the papers of Al-Najar (2013) and Goddard, Tavakoli, and Wilson (2005) and therefore will be introduced in the following way:

Null hypothesis (H0): The Economic Sanctions tend to have no effect on the cash holdings

(measured in cash ratio) on the companies, which were subject to the sanctions.

Alternative hypothesis (H1): The Economic Sanctions will have a positive effect on the cash

holdings (measured in cash ratio) of the companies, which were subject to the sanctions.

In addition, the following hypothesis concerning the profitability will be tested:

Null hypothesis (H0): The Economic sanctions will not affect the profitability (measured in

ROA) of the Russian firms

Alternative hypothesis (H1): The Economic sanctions will have a negative effect on the

profitability (measured in ROA) of the Russian firms

5. Methodology

The following methodology and all the estimation procedures may be used not only with economic sanctions but also for testing of the effect of other financial ratios on the dependent variables. The following will include the effects on the cash ratio, as a measurement of corporate cash holdings and the return on assets, which will stand as the measurement of profitability. Such method should help us not only to see the possible impact of the sanctions but also to see the estimated effect of the specific ratios or regressors on the cash holdings and profitability, for the model to fully explain the determinants.

(16)

Firstly, the descriptive statistics of financial parameters will be provided. Later on, it will be split into parts. The first descriptive statistics table is when the company does not bear any sanctions at a specific moment in time. The tables include the observations of the companies during which the sanctions were not implemented at all and the observations of the companies before the sanctions have been taking place. More about the data selection will follow in the ‘data’ part of the paper.

A Panel Data Difference in Difference (DID) approach will be used to estimate the effect of the specific financial parameters on cash ratio, which will be an indication for the firm cash holdings. Such regression will be similar to the previously mentioned regression in the paper of Al-Naijar (2013). Moreover, the specific variable SAN will be introduced. The variable is constructed deliberately to check for the actual effect of the economic sanctions on the Russian companies and also to see the other effects on the firm’s cash ratio. Therefore, the following regressions will be put into practice:

Regression 1.1 Estimating the cash ratio (CASH) using the random effects.

+

+

effects.

,

(17)

The following variables will be defined below:

Table 1. Definitions of the variables for the first regression

Variable

Definition

𝑪𝑨𝑺𝑯𝒊𝒕 A dependent variable, which stands for the cash ratio. It is the number of cash

holdings of the company divided by the total assets.

𝑳𝑬𝑽_𝒊𝒕 The leverage ratio, which can be defined as the ratio of total debt to total assets.

𝑹𝑶𝑬𝒊𝒕 The return on equity. The ratio of net income over total equity.

𝑫𝑷𝑶𝒊𝒕 Dividend payout ratio, which is calculated by diving dividends per share over

earnings per share. Note that such parameter cannot be found all the time, due to the fact that neither some of the studied companies will be private and not traded nor the full information disclosure will be available.

Time The dummy variable, which stands for the moment at which the sanction

implementation has already taken place. Namely, it will be equal to 1 if the period is after the third quarter of 2014 and zero otherwise.

Firm The dummy variable, which stands for whether each particular company has

ever faced sanctions. In the following research 12 companies have faced sanctions and 12 have not faced at all.

𝑺𝑨𝑵 The dummy interaction variable and stands for the case whether a particular company has faced the sanctions and at the specific time or not. Can also be defined as ‘Time× 𝐹𝑖𝑟𝑚’.

𝑳𝑰𝑸_𝒊𝒕 Liquidity ratio, which stands for the ratio of liquid assets over total liabilities.

F Firm Fixed Effects

T Time Fixed Effects

The following model will be tested for two fixed effects, namely for the company (firm) fixed effects and for the time fixed effects.

Furthermore, the second panel regression is introduced as well. Such model will be a tool to determine the effect of the economic sanctions on the firm’s profitability. The following

(18)

model takes some useful tools from the paper of Goddard, Tavaloki, and Wilson (2005), namely to use the return on assets as the dependent variable. Also, the ‘gearing variable’ of such authors was replaced by the ‘leverage ratio’, and the ‘liquidity ratio’ remains unchanged. It is the case that Goddard, Tavaloki and Wilson do not discuss sanctions. Therefore, the specific binary variables for the sanction definition are fully introduced by the author of this paper.

Such regression will be relatively similar in structure to the regression for the Cash ratio (‘CASH’). Although there are several crucial differences. The dependent variable will now be the return on assets (‘ROA’). The explanatory variables of ‘ROE’ and ‘DPO’ are no longer present. The reason for that is that ‘ROE’ and ‘ROA’ turn out to have a high correlation, so the estimation of one to another is not the best scenario. The ‘DPO’ variable also seems not to be a very good estimate. Moreover, another variable named as ‘CAP’ will be introduced as well. As a result, the following regression will help to test the second hypothesis. Therefore, such model will be formed in the following way:

_𝟒(𝟏)

𝑺𝑨𝑵 +

+

_𝟎(𝟒)

+ 𝜸

_𝟏(𝟒)

𝑳𝑰𝑸

_𝒊𝒕

+ 𝜸

_𝟐(𝟒)

𝑳𝑬𝑽

_𝒊𝒕

+ 𝜸

_𝟑(𝟒)

𝑪𝑨𝑷

_𝒊𝒕

+ 𝜸

_𝟒(𝟒)

𝑺𝑨𝑵

+ 𝑻 + 𝑭 + 𝜺

_𝒊𝒕

,

(19)

The definitions of the mentioned variables will be defined below:

Table 2. Definitions of the variables for the first regression

Variable

Definition

𝑹𝑶𝑨_𝒊𝒕 A dependent variable standing for the return on assets, which is typically specified. It stands for the ratio of net income over total assets.

𝑳𝑬𝑽_𝒊𝒕 The leverage ratio, which can be defined as the ratio of total debt to total assets.

𝑪𝑨𝑷_𝒊𝒕 Such variable stands for cash flow to capital expenditure ratio. It is defined with the ratio of cash flow from operating activities divided by capital expenditures.

Time The dummy variable, which stands for the moment at which the sanction

implementation has already taken place. Namely, it will be equal to 1 if the period is after the third quarter of 2014 and zero otherwise.

Firm The dummy variable, which stands for whether each particular company

has ever faced sanctions. In this work, 12 companies have faced sanctions and 12 have not faced at all.

𝑺𝑨𝑵 The dummy variable and stands for the case whether a particular company has faced the sanctions and at the specific time or not. 𝑳𝑰𝑸_𝒊𝒕 Liquidity ratio, which stands for the ratio of liquid assets over total

liabilities.

F Firm Fixed Effects

T Time Fixed Effects

The choice of using both random effects and fixed effects for the regressions is

straightforward. Once using the fixed effects method, it will be considered that the specific individual effects, such as the effects of time (Time Fixed Effects) and the effects of firms (Firm Fixed Effects) are correlated with the independent variables. Moreover, a Random Effects model is introduced for the reason that it is assumed that the specific individual effects (Time and Firm Fixed effects) are uncorrelated with the explanatory variables. The preferability of the methods will be later tested with the help of Durbin-Wu Hausman test. In case it will be proven that the

(20)

Random Effects model holds it will be considered more efficient. The following results will be displayed in the section 7.1.3 and the Durbin-Wu Hausman test will be shown in the section 7.3.

In addition, the robustness check in the section 7.2 will be performed. It will eventually be a useful tool to assess the reliability of the chosen model and evaluate whether this particular regression specification has been chosen correctly. For instance, in this section, the previously defined regressions with robust standard errors will be displayed. Such method will be done with the ‘vce (robust)’ option.

Important notice concerning the two above chosen models is that the market analysis of the stock prices and returns have not been taken into consideration. The following assumption has been taken because it is relatively ambiguous to evaluate the stock and market efficiency in regards to the economic sanctions imposed on Russia. Moreover, it is also the reason why the panel data regression has been used and not the event study method, which uses the abnormal returns of the chosen companies.

6. Data

The data collection starts with defining the Russian companies, who were influenced by the economic sanctions in 2014. The selection of such companies will be fully based on one official sanctioned statement, namely “Ukraine-related Sanctions; Publication of Executive Order 13662 Sectoral sanctions Identification list”. Such statement was made by the American Office of Foreign Assets Control on the 21st_{of July 2014, and it administers and enforces economic and trade}

matters. Such treaties are based on US foreign policy and national security goals and are under the supervision of the US Department of the Treasury.

Therefore, it was decided to collect the data for the 12 big Russian companies on which the sanctions were not imposed and compare it to the 12 major Russian firms, on which the sanctions were implemented. The extent of importance is based on the combination of the market capitalization and total revenues (sales). Some companies from the top-tier have not been taken into account, due to the insufficient amount of the information disclosure. If applicable, a strong emphasis on the companies in the financial, energy and military defense sectors was done.

As a result, the first group will be the control group and the second is the treatment one. Afterward, the historical balance sheets of such companies will used. This is done in order to collect all the relevant data, namely not only the cash holdings but also some specific financial indicators, such as cash, total assets, total equity, total debt, and dividends. This data was mostly

(21)

turned out to be digitally unavailable it was also extracted from the websites of such companies, namely from their financial reports. The time span will be defined, as the quarterly one. The starting period will be the first quarter of 2010, so it does include not only the sanctioned period, but also the four years prior to it. The last period taken will be the fourth quarter of 2017. As a result, eight years of data will be studied, in which all four quarters will be taken making an approximate amount of observations equal to 32.

Table 1. Expected relationship of independent variables on the variable

𝐂𝐀𝐒𝐇

:

Variable

Expected effect

Reasoning

LIQ Positive Liquid assets include cash in major therefore the estimated effect should be positive. This reasoning is slightly different from Al-Najar (2013) because he has separated cash and liquid assets.

LEV Positive An increase in leverage due to an increase in debt might have a positive effect on cash holdings. The companies might borrow money from other entities and keep such funds in cash. Moreover, according to Al-Najar (2013), the firms with higher leverage tend to have more cash due to the higher probability of financial distress.

ROE Positive An increase in return on equity should have a positive effect on the cash holdings in relationship to total assets. Such reasoning is based on the fact that the newly accumulated income might still be accumulated in the company and not reinvested. Such reasoning also coincides with Al-Najar, who says that return on equity is an outcome of the financing and investing activities. Therefore, profitable firms will have more cash in their stockpiles.

DPO Negative Firms should be less likely to hold more cash, given the fact the dividend payouts will increase. Furthermore,

(22)

Al-Najar also states that the dividend-paying firms can trade off the costs of holding cash by reducing dividend payments.

SAN Positive Companies may tend to hold more cash in relationship to total assets due to the precautionary motives. Such reasoning is especially valid in the banking industries

Table 2. Expected relationship of independent variables on the variable

𝐑𝐎𝐀:

Variable

Expected effect Reasoning

LIQ Positive Liquid assets may be easily reinvested, which might at some point in time bring higher net income in

relationship to total assets. Goddard, Tavaloki, and Wilson (2005) also state that an increase in liquidity reduces the risk exposure of not being able to maintain short-term financial commitments. The authors also report that the liquidity is an essential driver of corporate profitability.

LEV Negative An increase in leverage due to an increase in debt might have a negative effect on ROA. The companies might borrow money from other entities and will need to maintain the debt. Therefore, the companies will be making less profitable investment and therefore will be making fewer profits. Goddard, Tavaloki, and Wilson (2005) suggest that a high value of non-current

liabilities usually tends to decrease output and therefore have a negative effect on the profitability. Since long-term debt (part of the leverage numerator) is in the non-current liabilities the effect is also negative.

CAP Positive An increase in cash flow to capital expenditure ratio will imply that the company has no problems generating cash flows to acquire the asset acquisitions. Such asset

(23)

acquisitions may bring additional profits at some point in time.

SAN Negative An implementation of the economic sanctions may have a negative effect on the profitability of the companies. More reasons for that are explained in section 2.3

7. Results

This section covers the quantitative results of the study. First of all, the correlation tables and descriptive statistics will be provided. Second, the regression outcomes with the detailed explanations will be outlined. Furthermore, some robustness checks on the reliability of the results will be made. Lastly, the discussion and evaluation of the empirical results will be shown. The STATA software will be used since its implementation allows the readers to understand the actual meanings and estimation results.

7.1 Main results

The descriptive statistics including the binary variables is shown below. First of all, the correlation tables for the two regressions will be made.

7.1.1 Correlation tables

Table 3.1 Correlation table for the ‘CASH ratio’ (cash holdings)

regression

CASH LIQ ROE LEV SAN DPO Time Firm

CASH 1.0000 LIQ 0.3730 1.0000 ROE 0.1262 0.0999 1.0000 LEV -0.0552 -0.4345 -0.2227 1.0000 SAN -0.0331 -0.2544 -0.1027 0.0070 1.0000 DPO 0.1019 0.0072 -0.0164 0.0397 0.0036 1.0000 Time 0.0820 -0.0487 -0.0383 0.0115 0.6015 0.0588 1.0000 Firm -0.0272 -0.3693 -0.0120 -0.0083 0.5242 -0.0201 -0.0216 1.0000

(24)

Table 3.2 Correlation table for the ‘ROA’ (profitability) regression

ROA CAP SAN Firm Time LIQ LEV

ROA 1.0000 CAP 0.0319 1.0000 SAN -0.0021 0.0207 1.0000 Firm 0.0942 -0.0399 0.5254 1.0000 Time 0.0035 0.0254 0.5865 -0.0252 1.0000 LIQ 0.1769 0.0449 -0.2297 -0.3559 -0.0219 1.0000 LEV -0.3586 -0.0503 0.0001 -0.0121 0.0171 -0.4576 1.0000

By looking at the correlation tables for both of the regressions namely the regression for cash ratio and return on assets one can infer that there is not a single high correlation between any of the variables, if not including the relationship between the binary variables Firm and Time and its interaction term. The only relatively high negative correlation can be found between ‘LIQ’ and ‘LEV’, but it is still not significantly different from zero and can also be explained by the fact one the variables takes into account Liquid Assets (variable ‘LIQ’) in its numerator and the second one (‘LEV’) has to include total Assets in the denominator. Overall, one can conclude that since the correlations between variables are not high, it can be considered as a good sign. Therefore, further analysis can be done.

7.1.2 Descriptive statistics

Table 4.1 Descriptive Statistics in the Case of Firms not being influenced by

sanctions

Variable

Number of

observations

Mean

Standard

deviation

Min

Max

LIQ 523 .6414177 .5245315 .1054404 5.6426

ROE 525 .0474099 .1104837 -.2948815 1.057049

LEV 525 .2645984 .1520051 0 .7391765

DPO 557 19.46002 188.8744 0 4436.1

(25)

ROA 522 .0238862 .0428149 -.1763253 .4128759

CASH 552 .0859864 .0695084 .0036755 .5932362

Table 4.2 Descriptive Statistics in the Case of Firms being influenced by

sanctions

Variable

Number of

observations

Mean

Standard

deviation

Min

Max

LIQ 161 .3531962 .201583 .1104855 .7638786 ROE 161 .0188736 .0982594 -.5327243 .4356949 LEV 161 .2634329 .16337 .0153864 .7582438 DPO 158 20.53419 25.08492 0 116.109 CAP 138 -2.061518 53.17321 -309.5 356 ROA 161 .0208573 .0603271 -.0518992 .4128759 CASH 158 .0867865 .0518536 .0131909 .2401921

The following tables provide readers with a descriptive statistics comparison between the variables in the case there were no sanctions for the companies implemented at the specific point in time and the second table is the inverse, namely when the companies were under the effect of sanctions. It can already be seen several variables change quite a lot, once the companies become affected by the sanctions.

For instance, the LIQ parameter drops from 0.64 to 0.35, implying the fact that either the number of liquid assets in such companies go down or the amount of total liabilities in comparison to liquid assets increase. Moreover, the return on equity goes down significantly: from 0.044 to 0.018. The essential variables for this paper, which are CASH and ROE also differ and become lower and higher, respectively once the SAN binary variable becomes 1. In fact, there is no surprise in such changes. For instance, CASH ratio goes up due to the fact that it is frequently used as a precautionary motive against potential disturbances by many companies, and here I assume it to be the case as well. The return on assets decreased because the net income of such companies in comparison to total assets had been reduced. Such change seems logical enough, as it becomes harder to trade and provide services to the clients who have implemented such sanctions.

(26)

Table 4.3. Mean-comparison test (T-test).

Such test compares groups of data from the Table 4.1 (no Sanctions) to the groups of data from the Table 4.2 (Sanctions are present)

Variable Probability that the

Difference between groups and 1 and 2 is negative

(Ha: diff < 0)

Probability that the Difference between groups and 1 and 2 is

zero (Ha: diff != 0)

Probability that the Difference between groups and 1 and 2 is positive (Ha: diff > 0)

LIQ 1.0000 0.0000 0.0000 ROE 0.9983 0.0034 0.0017 LEV 0.5333 0.9334 0.4667 DPO 0.4716 0.9432 0.5284 CAP 0.3012 0.6024 0.6988 ROA 0.7601 0.4797 0.2399 CASH 0.4466 0.8932 0.5534

By looking at this table and evaluating the findings, it is important to mention that not all of the variables show that the difference between two groups (absence and presence of sanctions at a particular point in time) exists. For example, the CAP values tend to become higher when the economic sanctions are present, and this is in line with the initial expectations that were mentioned in Table 2 of Section 6. The values of ROA (Return on Assets) tend to become smaller (in line with the initial expectations and reasoning), the same can be said about ROE (Return on Equity) and LIQ (Liquidity Ratio). On the contrary, based on such difference test, there is no significant difference in observations between the observations for the DPO (Dividend Payout Ratio), LEV (Leverage Ratio) and CASH (Cash Ratio). It is important to mention that this test should not be considered as a crucial benchmark given the fact the relationship between the variables cannot be fully determined without an econometric model. Therefore, the analysis on the effect of the economic sanctions should be mostly based on the previously defined econometric models. The results of such models will follow in the next table of this paper.

(27)

7.1.3 Econometric regressions to answer the research question

Table 5. Panel data difference-in-difference regressions using CASH

(cash ratio) as a dependent variable.

In this section, the main regression outcomes for CASH will be displayed. The first column labels the variables that explain the dependent variable (CASH). The second column shows the outcomes for the Random Effects model (1). The third (2), fourth (3) and fifth (4) label the results for the Firm Fixed, Time Fixed, Time and Firm Fixed effects respectively. The table also reports the number of observations, the values of Wald chi2 test, the p-value of the Wald chi2 test and the values of rho. The standard errors are reported in parentheses and are homoscedastic. *, **, *** indicate significance at 10%, 5%, 1% respectively.

Dependent variable: CASH

(1) (2) (3) (4) Random Effects Firm Fixed Effects Time Fixed Effects

Time and Firm Fixed effects LIQ 0.0686*** 0.0691*** 0.0689*** 0.0701*** (0.00586) (0.00596) (0.00597) (0.00606) ROE 0.0427*** 0.0414*** 0.0387** 0.0359** (0.0151) (0.0151) (0.0154) (0.0153) LEV 0.140*** 0.146*** 0.153*** 0.168*** (0.0207) (0.0212) (0.0228) (0.0237) SAN -0.0171*** -0.0170*** -0.0174*** -0.0172*** (0.00513) (0.00513) (0.00519) (0.00512) DPO 2.44e-05*** 2.43e-05*** 2.68e-05*** 2.66e-05***

(7.44e-06) (7.44e-06) (7.68e-06) (7.58e-06) Constant -0.0119 -0.0362*** -0.0171 -0.0433*** (0.0190) (0.00959) (0.0180) (0.0131) Observations 671 671 671 671 Wald chi2 212.17 2174.32 243.08 2227.36 Prob > chi2 0.0000 0.0000 0.0000 0.0000 rho .74949453 0 .63499471 0

(28)

Time FE NO NO YES YES

The following table shows the four previously mentioned regression specifications where the variable CASH is the dependent variable.

First of all, it is important to look on the primary results of the specified model and the following regressions, namely on the effect of the economic sanctions (SAN) on the cash ratio (CASH), as it will help to answer the first hypothesis of the paper. It may be inferred that the effect of sanctions on the cash ratio is negative and significant for all of the specified regressions mentioned above. Moreover, it is in line with the initial expectations that were mentioned in Table 1 under the section ‘Data’. In all of the regressions, such effect is significantly different to the alpha of 1 percent, but the most substantial effects were captured under the regression (1), which stands for Random Effects and regression (4), which includes Time and Firm fixed effects. Once looking at the values of the standard errors for SAN one might notice that the highest standard error is when Time Fixed effects are included (regression (3)). On the contrary, the above regressions do not indicate very different results for the effect of sanctions on the cash ratio.

If it comes to evaluate the other effects on the cash ratio, which might also be interesting to look at it should be first said that most of the above estimates are strongly significant. For instance, one might notice that the effect of return on equity (ROE) on the cash ratio (CASH) is positive and it coincides with the initial expectations mentioned in the data section 6. It is strongly significant (to the alpha of 1 percent) in the case of Random and Firm Fixed effects and significant to 5 percent in the case of regressions (3) and (4). The relationship of liquidity ratio (LIQ) on CASH is as expected, namely positive and strongly significant in all 4 regression specifications. The same case applies to the effect of LEV (leverage ratio) on CASH. On the other hand, when it comes to evaluating the effect of the dividend payout ratio (DPO) on CASH it is not as it was initially expected in section 6 and such effect is still strongly significantly different from zero.

Then it comes to look at the values of the Wald chi2 test and its p-values, which tests whether the explanatory variables in the specified model are different from zero. Once evaluating the actual values of the test and looking at the p-values it should be said that since the values of such test statistic is rather prominent in all regression specifications it may be concluded that there is no possibility that at least one explanatory variable has a value of zero and as a result this might be another indication that such explanatory (independent) variables have been chosen correctly. Last but not least it is also interesting to look at the values of rho, which are not all always the same. For instance, in the case of Random (1) and Time Fixed (3) Effects, they are relatively big, while in the other regressions it is zero.

(29)

Table 6. Panel difference-in-difference regressions using ROA (return on

assets) as a dependent variable.

In this section, the main regression outcomes for ROA will be displayed. The first column labels the variables that explain the dependent variable (ROA). The second column shows the outcomes for the Random Effects model (1). The third (2), fourth (3) and fifth (4) label the results for the Firm Fixed, Time Fixed, Time and Firm Fixed effects respectively. The table also reports the number of observations, the values of Wald chi2 test, the p-value of the Wald chi2 test and the values of rho. The standard errors are reported in parentheses and are homoscedastic. *, **, *** indicate significance at 10%, 5%, 1% respectively.

Dependent variable: ROA

Time and Firm Fixed Effects

CAP 6.60e-06 5.39e-06 2.05e-05 1.46e-05

(5.45e-05) (5.46e-05) (5.86e-05) (5.65e-05)

SAN -0.0208*** -0.0209*** -0.0208*** -0.0211*** (0.00592) (0.00593) (0.00620) (0.00599) (0.00414) (0.00414) (0.0144) (0.0139) LIQ 0.0119* 0.0134** 0.00855 0.0124* (0.00642) (0.00680) (0.00602) (0.00700) LEV -0.0898*** -0.0815*** -0.104*** -0.0798*** (0.0229) (0.0252) (0.0209) (0.0283) Constant 0.0300** 0.0404*** 0.0381*** 0.0430*** (0.0137) (0.0111) (0.0144) (0.0152) Observations 638 638 638 638 Wald chi2 212.17 2174.32 243.08 2227.36 Prob > chi2 0.0000 0.0000 0.0000 0.0000 rho .74949453 0 .63499471 0

Firm FE NO YES NO YES

Once looking at the second panel difference-in-difference table, where the independent variables (SAN and others) are regressed on the ROA (return on assets) it is first of all interesting

(30)

second hypothesis of the paper. It may be outlined that the effect of sanctions on the return on assets is negative and it is significantly different from zero at almost any level (it is significant at one percent). Such relationship eventually brings no surprise, as it is utterly in line with the expectations under the section 6 of the paper. Once zooming into more details of the found estimates to evaluate the differences between the four regression specifications (1-4) it should be mentioned that the most potent negative effect was in the specification was in the case when the Time and Fixed effects specification were assumed. Although, the standard error for such estimate is also not the smallest, but not the largest either. In conclusion, it is the Random (1) and Time and Firm Fixed (4) effects to show the strongest negative impact of the economic sanctions on the Return on Assets, which is a measurement of profitability.

Once evaluating the other explanatory variables of the above model, it can be said that the effect of leverage (LEV) is negative and strongly significantly different from zero and it is in line with the expectations under the ‘Data Section’. On the other hand, even though the effect of liquidity (LIQ) on the ROA is positive, as expected it is not strongly significant. For instance, an average significance (to 5 percent) can be found when only the Time Fixed Effects (2) are included, and a weak significance (to 10 percent) can be inferred in the case of Random (1) and Firm and Time Fixed (4) effects. In the case of Firm Fixed Effects (3), there is no significance what so ever. Furthermore, the impact of CAP cannot be precisely defined as it is not significantly different at any level in all regression specifications.

The values of Wald Chi2 test are relatively solid, and therefore the p-values are very small. As a result, this is an indication that a hypothesis that at least one of the estimates in such model is equal to zero is rejected. Such case may infer that the estimates can be considered as relatively good and strong.

7.2 Additional results (robustness checks)

In order to assess the validity of the previously performed tests, it will be useful to perform several robustness checks. For instance, the previously selected models with the robust standard errors options will be used. Moreover, the Durbin-Wu Hausman test will be performed. Such test will be helpful to understand whether it is better to pay attention to the results of the Random Effects models or to the particular test with the selected fixed effects.

(31)

7.2.1 The robust standard errors usage

Table 7. Panel difference-in-difference regressions using CASH (cash ratio)

as a dependent variable with the robust standard errors.

In this section, the main regression outcomes for ROA will be displayed. The following test is conducted with the robust standard errors. The first column labels the variables that explain the dependent variable (ROA). The second column shows the outcomes for the Random Effects model (1). The third (2), fourth (3) and fifth (4) label the results for the Firm Fixed, Time Fixed, Time and Firm Fixed effects respectively. The table also reports the number of observations, the values of Wald chi2 test, the p-value of the Wald chi2 test and the values of rho. The standard errors are reported in parentheses and are robust (heteroscedastic). *, **, *** indicate significance at 10%, 5%, 1% respectively.

Dependent variable: CASH

Time and Firm Fixed effects LIQ 0.0686*** 0.0691*** 0.0689*** 0.0701*** (0.00832) (0.00832) (0.00898) (0.00889) ROE 0.0427** 0.0414** 0.0387** 0.0359** (0.0172) (0.0165) (0.0179) (0.0159) LEV 0.140** 0.146* 0.153** 0.168** (0.0712) (0.0750) (0.0703) (0.0763) SAN -0.0171 -0.0170 -0.0174 -0.0172 (0.0106) (0.0107) (0.0108) (0.0110) DPO 2.44e-05*** 2.43e-05*** 2.68e-05*** 2.66e-05***

(2.37e-06) (2.47e-06) (1.98e-06) (2.10e-06)

Constant -0.0119 -0.0362 -0.0171 -0.0433* (0.0235) (0.0244) (0.0216) (0.0235) Observations 671 671 671 671 Wald chi2 3552.93 . . . Prob > chi2 0.0000 . . . rho .74949453 0 .63499471 0

(32)

Such Table 7 is relatively similar to the Table 5 with one difference, namely that the standard errors are robust (non-constant) in the case of heteroskedasticity. Once using such option, it may be said that most of the estimates tend to show an insignificant relationship and the effect of Sanctions (SAN), which is the most crucial variable in the model is not an exception. Therefore, one can conclude that the regression specification under Table 5 is correct in comparison to Table 7, as it gives better and more accurate estimation outcomes. As a result, it may be concluded that the initial specifications have been chosen right.

Table 8. Panel difference-in-difference regressions using ROA (return on

assets) as a dependent variable with the robust standard errors.

In this section, the main regression outcomes for ROA will be displayed. The following test is conducted with the robust standard errors. The first column labels the variables that explain the dependent variable (ROA). The second column shows the outcomes for the Random Effects model (1). The third (2), fourth (3) and fifth (4) label the results for the Firm Fixed, Time Fixed, Time and Firm Fixed effects respectively. The table also reports the number of observations, the values of Wald chi2 test, the p-value of the Wald chi2 test and the values of rho. The standard errors are reported in parentheses and are heteroscedastic. *, **, *** indicate significance at 10%, 5%, 1% respectively.

Dependent variable: ROA

Time and Firm Fixed Effects

CAP 6.60e-06 5.39e-06 2.05e-05 1.46e-05

(5.20e-06) (5.05e-06) (1.81e-05) (1.68e-05)

SAN -0.0208*** -0.0209*** -0.0208*** -0.0211***

(0.00479) (0.00483) (0.00502) (0.00486)

LIQ 0.0119 0.0134 0.00855 0.0124

(33)

The above Table 8 is similar to the Table 6 but uses an assumption of robust standard errors. An outcome of such specification is similar, as it was shown under the explanations under Table 7 when the heteroskedastic models in in such table were compared to the homoscedastic ones in Table 5. In more details, even though the effect of SAN (sanctions) on the ROE (return on equity) is still positive and significant the same about the other variables cannot be said. Such findings in combination with the indefinite values of Wald chi2 is the sign that the initial model in Table 6 is more accurate.

7.3 Durbin-Wu Hausman tests

Table 9.1 Durbin-Wu Hausman test for the choice of firm fixed effects versus

random effects for the model 1 (Cash Ratio)

VARIABLES Firm Fixed Effects Random Effects (b-B) Difference sqrt(diag(V_b-V_B)) S.E. LIQ .0691162 .0685771 .0005391 .0010472 ROE .0414388 .0426717 -.0012329 .0008477 LEV .1464766 .1404686 .006008 .0046585 SAN -.0169898 -.017093 .0001031 .0001336

DPO .0000243 .0000244 -1.51e-07 1.71e-07

b = consistent under Ho and Ha; obtained from xtreg

B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic

LEV -0.0898*** -0.0815*** -0.104* -0.0798* (0.0282) (0.0290) (0.0602) (0.0479) Constant 0.0300* 0.0404*** 0.0381* 0.0430** (0.0154) (0.0129) (0.0220) (0.0167) Observations 638 638 638 638 Wald chi2 40.99 . . . Prob > chi2 0.0000 . . . rho .42929235 0 .15065307 0

(34)

chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 1.43

Prob>chi2 = 0.9640

In this table, a small robustness check was done to test whether the firm fixed effects model fits better in comparison with the standard random effect model for the regression of CASH. For this case, the Durbin-Wu Hausman test for the choice of firm fixed effects versus random effects was performed. The null hypothesis states that the difference between the coefficients of the two models is not systematic and the alternative states the opposite. In such test, it is essential to look at the value of ‘chi2(4)’ and its p-value. Since the p-value is relatively high, it may be outlined that the null hypothesis is not rejected. Moreover, since the difference in the coefficients turns out not to be systematic, both of the regression specifications can be used efficiently and effectively. Therefore, it turns out to believe that the firm fixed effects model is as good as the random effects model.

Table 9.2 Durbin-Wu Hausman test for the choice of time fixed effects versus

random effects for the model 1 (Cash Ratio)

VARIABLES Time Fixed Effects Random Effects (b-B) Difference sqrt(diag(V_b-V_B)) S.E. LIQ .06893 .0685771 .0003529 .0007891 ROE .0387491 .0426717 -.0039226 .0026151 LEV .1525481 .1404686 .0120795 .0090122 SAN -.017417 -.017093 -.000324 .0003323

DPO .0000268 .0000244 2.33e-06 1.59e-06

chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 10.29

(35)

In this table, a Durbin-Wu Hausman test to compare time fixed effects and random effects is conducted, and it is similar to the test in table 9.1. In the case of such comparison, the similar reasoning as in the analysis above can be done. Since the p-value of chi2 in this table is above 10 % it is concluded that the difference between the models is not systematic and

similarly, the time fixed effects should be as favored as the random effects.

Table 9.3 Durbin-Wu Hausman test for the choice of time and firm fixed

effects versus random effects for the model 1 (Cash Ratio)

VARIABLES Time and Firm Fixed effects Random Effects (b-B) Difference sqrt(diag(V_b-V_B)) S.E. LIQ .070068 .0685771 .0014909 .0016199 ROE .0359078 .0426717 -.0067639 .0028893 LEV .1682574 .1404686 .0277888 .0118125 SAN -.0172031 -.017093 -.0001101 .0003806

DPO .0000266 .0000244 2.17e-06 1.60e-06

chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 3.66

Prob>chi2 = 0.7230

In the Table 9.3, the similar Durbin-Wu Hausman test as in the previous cases was performed. The results of such test outline the same tendency as in the last test: null hypothesis cannot be rejected and a preferable model for such regression specifications cannot be defined.

The effect of economic sanctions on corporate cash holdings and profitability : the case of Russia

Master Thesis