The Impact Of Penny Stocks On The Pricing Of Companies Listed On The Warsaw Stock Exchange In Light Of The CAPM

Abstract Oryginality and objective – Research on the pricing of stocks listed on developed markets shows inexplicable deviation from the pricing that could be observed with CAPM validity. A similar anomaly is found on the Polish market. Reasons for inconsistent pricing with CAPM are unknown, and they are the main objective of this research. Method – The study is conducted using stocks listed on the Warsaw Stock Exchange in 1995–2012. Quintile stock portfolios are formed on the basis of strategies widely used by investors. The study is carried out in several modes. In the subsequent modes penny stocks with the market values below 0.5, 1.5, 5.0 and 15.0 PLN are eliminated. Results – It is conjectured that both penny stocks and improper procedures for the test portfolios forming contribute to inconsistent stock pricing in light of the CAPM. The studies show that results are in line with the extended conjectures. Also, study results indicate that speculative stocks are mostly penny stocks, however, it is not possible to explicitly state that penny stock are speculative.


Introduction
The classic CAPM is commonly used to assess the capital cost of companies listed on the stock exchange. In this case, the main condition for a correct estimation of capital cost is the assumption that stock pricing is consistent with the pricing that could be observed with CAPM validity. Fama  The attempt to explain the reasons for the incorrect stock pricing in light of the CAPM is undertaken by Urbański 13 . He analyzed the quarterly returns of stocks listed on the Warsaw Stock Exchange (WSE) in 1995-2012, and notes that many speculative stocks, described by bad financial indicators and penny prices, are characterized by extremely high returns. The author of this work examines the impact of speculative stocks on the pricing and states that they are components of inconsistent stock pricing in light of the CAPM.
It seems commonly understood that investment in penny stock is characterized by high risk arising from simple speculative reasons. Consequently, a large number of listed penny stock makes the market risky and speculative. The most famous stock exchanges introduce restrictions on penny stocks trading 14 .
One might ask whether the penny stocks can be identified with speculative ones and vice versa, whether speculative stocks can be equated with the penny stocks. Following this thought further, in the light of Urbański's previous research, the question arises whether inconsistent pricing in light of the CAPM is a component of speculative stocks or penny stocks.
To our knowledge, there are no in-depth studies which explain the reasons for the incorrect stock pricing in light of the CAPM. In the paper we continue these works and study the impact of penny stock on pricing that could be observed with classic CAPM validity. Because the basis of proper testing of CAPM application is appropriate forming of portfolios, we apply two portfolio forming procedures, using the instructions proposed by Cochrane 15 . We analyze four modes of penny stock and two modes of speculative stocks influence for each portfolio building procedure. In addition, we explore the relationship between penny and speculative stocks. Therefore, we expect that the following conjectures are true:

Conjecture 1.
Penny stocks are speculative, characterized by weak financial indicators and high returns.

Conjecture 2.
Penny stocks are the components of inconsistent stock pricing in light of the CAPM.

Conjecture 3.
Improper procedures of portfolio forming lead to incompatible stock pricing.
Section 1 presents the fundamental model of portfolio management. Section 2 discusses the procedures of chosen methods of portfolio construction. In section 3 we study the relationship between penny stock and speculative stock. Section 4 widely analyzes the results of pricing in light of the CAPM for each case presented in Section 2. The final part of the paper presents conclusions.

The procedures of portfolio management
We analyze two procedures of portfolio construction. Procedure 1 is proposed by Urbański 16 . This model selects components of portfolio on the basis of the functional FUN, defined by equations (1), (2) and (3). This procedure of portfolio management provides practical investment strategies.
nor ROE nor AS nor APO nor APN FUN nor MV E nor MV BV Functions F j (j = 1, …, 6) are transformed to normalized areas <a j ; b j >, according to equation (3): In equations (1), (2) and (3), the corresponding indications are as follows: ROE is return on book equity; are values that are accumulated from the beginning of the year as net sales revenue (S), operating profit (PO) and net profit (PN) at the end of "i" quarter (Q i ); In procedure 2, portfolios are built according to Fama and French 19 methods. In this case portfolios are formed on the basis of BV/MV and capitalization (company size). The results of many previous works show a significant influence of BV/MV and capitalization on future returns 20 . The reason for these connections is explained by Fama and French 21 . This work shows that stock prices in relation to capitalization and BV/MV are influenced by earning structure in the last five years. Thus, one can conclude that prices (and returns) are directly generated by earnings in the period preceding the investment, while BV/MV and size also result from the recorded changes of earnings. Also, the above analysis indicates that portfolio forming parameters should be based on a company's earning structure in the last several years. These findings are taken into account by Urbański's model which (in the light of Fama and French 22 research) can be an alternative to Fama and French 23 method of portfolio construction.

Data and construction of testing portfolios
We analyzed the quarterly returns of the stocks listed on the WSE in 1995-2012 24 0.00 r is average spread value; * H 0 :`r = 0,H 1 : `r ≠ 0; ** H 0 : `rprocedure_1 = r procedure_2 , H 1 : r procedure_1 > r procedure_2 . In  The spreads for portfolios formed on FUN, NUM and DEN (in procedure 1) are significantly different from zero (p-values < 0.00). The spreads for portfolios formed on BV/MV and CAP (in procedure 2) are equal to zero. The spreads for portfolios formed in procedure 1 are significantly higher than for portfolios formed in procedure 2 (p-values < 0.01). However, the spreads for portfolios formed on FUN, NUM , DEN, BV/MV or CAP, but in different modes, do not differ 26 .      The largest number of speculative stocks is in the range of market values (1.00 PLN; 2.00 PLN>. In the case of speculative stocks S2 it is 238 stocks, which account for 11.73% of all S2 stocks. In the case of speculative stocks S1 it is 95 stocks, which account for 12.75% of all S1 stocks 27 . On the other hand, 49.58% of S2 and 51.54% speculative stocks S1 have a price of less than 5.00 PLN. These results indicate that speculative stocks are mostly penny stocks.

Penny stocks versus speculative stocks
However, in terms of conducted research, it is not possible explicitly to state that penny stock portfolios are speculative (see Figure 1 and Table 2). Stock market value, MV percentage of speculation stocks S2 with market value < MV+1 PLN percentage of speculation stocks S1 with market value < MV + 1 PLN percentage of speculation stocks S2 in range <MV; MV + 1 PLN) The percentage of speculation stocks is determined by the relation between the number of speculation stocks with market value < MV and the number of all speculation stocks, listed on WSE, at the beginning of 64 quarterly investment periods in 1996-2011. Speculative stocks S1 are defined as meeting one of the following boundary conditions: a) MV/BV > 100, b) ROE < 0 and BV > 0 and r it > 0, c) MV/BV > 30 and r it > 0, where ROE is the return on book value (BV), r it is the return of portfolio i in period t. Speculative stocks S2 are defined as meeting an additional condition d) MV/E < 0 and BV > 0, where E is the average earning for the last four quarters.

Stock pricing in light of the CAPM
The statistical model which tests the classic CAPM can be described by equations (4) and (5). The regressions of time series (4) are analyzed in the first pass. The equation (5) is analyzed in the second pass as the time-cross-section regression, using panel data.
The response variable of the above regressions is the excess of return  Table 4 presents the values of estimated parameters of regressions (5) and the crosssectional R 2 LL measure amployed by Lettau and Ludvigson (2001) 30 . If portfolios are built on FUN, NUM and DEN, and speculative stocks are removed from analysis (modes MS1 and MS2) the tested application prices the risk premium (γ M ) at the level of 12% and 19% quarterly. Also, the risk premium is priced if penny stocks with market values below 15.00 PLN (mode MP4) are not considered. In this case the value of γ M is similar to mode MS2 and equals 13%.
The elimination of stocks below 0.50, 1.50 and 5.00 PLN (modes MP1, MP2 and MP3) does not allow for significant estimates of the risk premium.
If portfolios are built on BV/MV and CAP, the classic CAPM does not price risk premium.
In this case, the values of γ M are insignificantly different from zero for all modes (p-values > 0.46).
Coefficient R 2 LL assumes extremely small values, in the cases of M1, MP1, MP2 and MP3 modes, and grows after elimination of stocks below 15.00 PLN, as well as exclusion of speculative stocks (in modes MS1 and MS2) assuming 23%, 11% and 57%, respectively.   Pricing errors decrease rapidly after elimination of speculative stocks. Also, the removal of penny stocks, but only with the market values below 15.00 PLN decreases pricing errors, at 0.08 significance level. It is documented by the values of Q A (F) statistic (see Table 5). This proves that mean-variance-efficient portfolios are generated if speculative stocks are excluded from consideration, while the removal of penny stocks with the market values below 5.00 PLN does not affect the portfolio efficiency. .00 PLN, respectively. Mode MS1 eliminates speculative stocks meeting one of the following boundary conditions: a) MV/BV > 100, b) ROE < 0 and BV > 0 and r it > 0, c) MV/BV > 30 and r it > 0, where P is the stock market value, ROE is the return on book value (BV), r it is the return of portfolio i in period t. Mode MS2 eliminates speculative stocks meeting additional condition d) P/E < 0, where E is the average earning for the last four quarters. RF is the 91-day Treasury bill return. M i, β is the loading on the market factor (RM -RF, for i portfolio) estimated from first-pass time-series regressions. RM is evaluated by the return on the WIG index of the WSE. GRS is the F-statistics of Gibbons et al. (1989). Q A (F) reports F-statistic and its corresponding p-value indicated below for Shanken's (1985) test that the pricing errors in the model are jointly zero. The response variable is excess return on 15 stock portfolios formed on FUN, NUM and DEN value in period t. The Prais-Winsten algorithm is used for correction of autocorrelation. The sample period is from 1995 to 2012, 64 Quarters.

Conclusions
In this paper we explore the impact of penny stocks on the pricing which would result from the correctness of CAPM assumptions. The conducted research leads to the following conclusions: 1. The return spreads for portfolios formed on FUN, NUM and DEN are significantly higher than spreads for portfolios formed on BV/MV and CAP, however the penny stocks do not affect the size of spread. 4. Speculative stocks are mostly penny stocks. However, in terms of conducted research, it cannot be explicitly stated that penny stocks are speculative. These results are not in line with Conjecture 1. 5. A systematic risk is significantly different from zero for all the tested cases, and it is similar for different modes and procedures of portfolio construction.
6. If portfolios are formed on FUN, NUM and DEN, the exclusion of stocks with market values below 5.00 PLN does not allow for significant estimates of the risk premium, however the risk premium is priced if penny stocks below 15.00 PLN are not considered.
7. If portfolios are built on BV/MV and CAP, the classic CAPM does not price the risk premium on WSE. Incorrect pricing is caused by improper procedures for the portfolio forming, characterized by small return spreads. This is in line with Conjecture 3.
8. If penny stocks (below 15 PLN) are excluded from portfolios, R 2 LL grows from 1% to 23%. This is in line with Conjecture 2. If speculative stocks are excluded from the portfolios, R 2 LL grows to 56%. 9. The removal of penny stocks from portfolios does not affect the intercepts values of regressions. This does not confirm Conjecture 2. If speculative stocks are eliminated values of intercepts fall to zero.
10. The removal of penny stocks, but only with the market values below 15.00 PLN, decreases pricing errors. This is in line with Conjecture 2. Also, pricing errors decrease after elimination of speculative stocks.
11. The removal of penny stocks with the market values below 5.00 PLN does not affect the portfolio efficiency. This does not confirm Conjecture 2. Classic CAPM generates mean-variance-efficient portfolios if speculative stocks are excluded from consideration.
The identification of correlations between penny stocks and speculation stocks requires further research. 1 Fama, MacBeth (1973).