AbstractThe low-frequency liquidity measures are accepted and widely used by academic community and practitioners. This paper replicates the concept of low-frequency liquidity measures from Johann and Theissen (2017). The empirical results from Thai equity market align with Johann and Theissen (2017) study and previous literatures. Contributions of the study will benefit, in term of liquidity assessment, to academic community and practitioner in Thailand.1. Introduction and motivationLiquidity is a complex concept. In general, liquidity often defined as the ability of markets to absorb large transactions without much effect on prices. Nowadays, research in liquidity is important for empirical asset pricing, market efficiency, and corporate finance literature. Especially, in market microstructure, liquidity plays a central role in the functioning of financial markets. A number of studies have proposed liquidity measures derived from intraday data, which we called “high-frequency data”. However, intraday data are not available in many countries. Even if the data is available, estimation of liquidity required high performance of computer and computational intensive process. Recently, Johann and Theissen (2017) perform comprehensive comparative analysis of low-frequency measures of liquidity by using US data. They identified high quality proxies for liquidity based on daily data. Hence, this study will use the low-frequency proxies to explore liquidity in emerging market of Thailand.The aim of this study is to provide empirical evidence about liquidity in Thai equity market in several perspectives, both cross-sectional determinants of liquidity and time-series variation in liquidity analysis. This study contributes to market microstructure literature in several ways. First, this study provides empirical evidence on liquidity in stock exchange of Thailand, before and after crisis. Second, this paper is the first study investigating low-frequency liquidity measurements in developing market of Thailand. The remainder of the paper is organized as follows. In Section 2, reviews the literature on liquidity and discusses evidence on low-frequency measures. Section 3, describe data structure and methodology for this study. Section 4, present the results. Last section, discuss the results and further enhancement. 2. Liquidity measures and literature reviewThe literature has used an extensive set of measures and proxies to estimate the liquidity on the stock market. The two most widely-used measures in liquidity are spread and price impact. Brennan and Subrahmanyam (1996) suggest that spread and price impact represent the fixed and variable components of the trading cost, respectively. For low-frequency measures of liquidity, Johann and Theissen (2017) present some low-frequency measures has high correlation with the benchmark measure. In this study, the data that available are daily close price, high price, low price, and traded volume. Due to data availability, I selected VoV Sigma, Amihud (2002) illiquidity ratio, and VoV daily which represent liquidity of spread and price impact. Moreover, the study of Johann and Theissen (2017) also show that these set of low-frequency measures have performed well along with the benchmark measure.2.1 Low-frequency estimators of the Bid-Ask spreadTobek (2016) develops modified versions of several low frequency estimators. He shows empirically that the bid-ask spread is closely related to a function of volume and volatility. The ratio will be: ? is estimated by the sum of squared daily returns, and Vi is the average of the daily trading volume.2.2 Low-frequency estimators of the price impactAmihud (2002) develops a price impact measure which represents the daily price response associated with one dollar of trading volume. Specifically, he uses the following ratio: ri,t and Vi,t are the return and the dollar trading volume of stock i on day t, respectively, and Di is the number of days in the evaluation period. Tobek (2016) also proposed a modified version of the Amihud (2002) illiquidity ratio based on the volatility-to-volume ratio. The daily version is defined as: Monthly and yearly estimates are obtained by averaging over the daily values.3. Data and Methodology3.1 DataIn order to calculate low-frequency measures of liquidity. I obtained stock data and market data from Datastream (Thomson Reuters). The data contains the daily price, adjusted for dividends and capital changes of all common stock listed on the Stock Exchange of Thailand (SET) during the period from January 2005 to December 2017, 13 years in total. This sample is representative of the market and cover the period of financial crisis in Thailand. In this study, stocks were separated into 2 groups, large and small firms. I used market capitalization as of December 2017 to identify size of firms. Data were sorted by market capitalization and adopted percentile method, 50/50 for large and small firms. SAS and Microsoft Excel software are main tools for data cleansing and statistical analysis.3.2 Methodology I started by computing low-frequency measures for each stock using formula as mentioned above. Afterward, I aggregated the liquidity across all stocks and then make analysis of the time-series variation in liquidity. I also presented the change in liquidity during the financial crisis. For cross-sectional analysis of liquidity, I performed descriptive statistics to show the difference in liquidity between large and small firms. However, I didn’t perform panel regression across determinants of liquidity because the limitation of the data. All low-frequency measures were calculated on monthly basis.4. ResultsTable 1 gives the number of firms analyzed, summarizes of market capitalization, and SET index during 2005-2017.Table 1 Descriptive summary of Thai equity marketYear Number of listed firms Market capitalization (Million Baht) SET Index 2005 368 4,437,047 713.732006 385 4,424,543 679.842007 398 5,886,495 858.102008 412 3,289,500 449.962009 423 5,404,929 734.542010 429 7,727,036 1032.762011 439 8,138,493 1025.322012 453 11,433,450 1391.932013 475 11,080,689 1298.712014 499 13,308,452 1497.672015 529 11,864,385 1288.022016 542 14,810,962 1542.942017 573 17,551,961 1753.714.1 Time-series variation in liquidityFigure 1 presents change in liquidity overtime. Particularly, during the financial crisis in year 2008-2010. Subprime mortgage crisis globally impacts to all equity market around the world including Thailand. During that period, SET index was dropped from 800 to 400 points. The liquidity significantly declined from the previous period. All low-frequency liquidity measures (Amihud, VoV daily, VoV Sigma) confirm the same trend of illiquidity. However, after the crisis, the liquidity in stock market revert to the same level as the period before crisis. Figure 1 Variation in liquidity over time. Not only for financial crisis that can impact to liquidity of equity market. Special events, such as political event in 2014, can be also impact to the liquidity of marker as well. Figure 2 shows you the movement of SET index during past 13 years along with some major events that impact the market down. Figure 2 SET index movement in last 13 years. The empirical results align with many literatures about stock market liquidity and its relation to financial crises. Moreover, low-frequency liquidity measures (Amihud, VoV daily, VoV Sigma) can well explain the liquidity in Thai equity market.4.2 Cross-sectional analysis of liquidityI separated firms into large and small firms by using firm’s market capitalization. Adopted percentile 50/50 for cut-off point. Table 2 gives the number of firms analyzed in each size, summarizes of market capitalization, and average market capitalization.Table 2 Descriptive summary of firm size in Thai equity marketSummary statistics Large firms Small firms Number of firms (N) 287 286Total market capitalization (Million baht) 16,976,635 575,325Average market capitalization (Million baht) 59,152 2,012Table 2 shows skewness in market capitalization between large and small firms. Next, I performed low-frequency liquidity measures (Amihud, VoV daily, VoV Sigma) both large and small firms for comparable. The results shown in figure 3-5. Figure 3 Liquidity for large and small firms using VoV Sigma. Figure 4 Liquidity for large and small firms using Amihud (2002) illiquidity ratio. Figure 5 Liquidity for large and small firms using VoV Daily. Using low-frequency liquidity measures, Large firms are more liquid than small firms. The result confirm consistency in all liquidity measures (Amihud, VoV daily, VoV Sigma). Particularly, the severity of illiquidity for small firms is extremely high during the crisis.5. DiscussionLiquidity plays an increasingly important role in empirical asset pricing, market efficiency, and corporate finance. This paper replicates low-frequency liquidity measures of Johann and Theissen (2017), which widely used by academic community and practitioners. The results show that low-frequency liquidity measures (Amihud, VoV daily, VoV Sigma) can well explain the liquidity in Thai equity market both time-series and cross-sectional analysis, consistency with Johann and Theissen (2017) study. The empirical show that large firms are the more liquid than small firms. The severity of illiquidity for small firms is extremely high during the crisis.There are still some limitations. Several aspects have yet to be investigated, to have more accurate results. The recommendations for further study, I suggest that further study should include following:? Investigate the result of low-frequency liquidity measures by comparing with high-frequency data (intraday data) for Thai equity market.? Investigate liquidity result by using low-frequency measures in other stock exchange (i.e. CLMV, Asia or Middle-east or China) to see whether low-frequency liquidity measures still perform well to explain the stock liquidity. ? For Cross-sectional analysis of liquidity, may use other technique such as panel regression and also looking for other determinants of liquidity.? Research for other method to separate size of firms.Future research to address this issue seems warranted.