Lot Measure and Zero Returns
Lesmond, Ogden, and Trzcinka (1999) (LOT) create a measure for liquidity based on zero returns. A zero return is when the price of a security does not change over a given time period. They look at the proportion of zero return days over a year and compare this measure to other proxies for liquidity such as, firm size and transaction costs. They find zero returns are inversely related to firm size and act as a proxy for transaction costs. The intuition behind zero returns is if no trade happens it must be that the benefit of trading for the marginal investor does not outweigh the cost of trading. The lower the costs of trading, which include bid-ask spread, commissions, expected price impact and opportunity costs the more liquid the asset. The measure was created specifically for situations where there is not the necessary information available for calculating the bid-ask spread such as, the corporate and municipal bond markets.
Chen, Lesmond, and Wei (2007) apply the LOT measure to the corporate bond market and find that the liquidity component of the yield spread is not directly related to default risk. Also by comparing the Lot measure with the bid-ask spread they come
to the conclusion that the measure can be used in liquidity studies where information is sparse.
Bekaert, Harvey, and Lundblad (2007) use the proportion of zero returns to proxy for illiquidity in emerging markets where data is sparse. They also create a price pressure measure, which tries to account for consecutive non-trading days, where more days in a row of non-trading may signify a more illiquid bond. They take the sum of the weighted returns of an index, including only stocks with zero returns, where the weights used are the capitalization weights of the stocks and the returns are the returns if the stocks had traded. They divide this sum by the sum of the weighted returns of all stocks in the index. They find this measure is highly correlated with the zero returns measure and therefore a good proxy for liquidity. This measure poses another possibility of measuring the liquidity e↵ect of insurance in the municipal bond market. However, because the municipal bond market is an over the counter market the only available data is for transactions that occur. I am unable to calculate zero return days as I do not know if the value changed or if there was simply no trade. This is why I am unable to use the two previous measures. Bid-Ask Spread
One of the most common measures of liquidity is the bid-ask spread. The smaller the spread the lower the price impact from selling a security and therefore the more liquid the security. Longsta↵, Mithal, and Neis (2005) are able to separate out the default and non-default component of corporate bonds using cds prices. By regressing the non-default component on measures of liquidity such as the bid-ask spread and principal amount outstanding. They conclude that the non-default component is strongly related to these liquidity measures.
Harris and Piwowar (2006) and Green, Hollifield, and Schrho↵ (2007) examine liquidity as measured through transaction costs in the municipal bond market. Harris and Piwowar (2006) estimate secondary trading costs and analyze the factors that a↵ect those costs. They improve upon the earlier literature by taking advantage of the MSRB data set, that I use, to incorporate information from every transaction. They improve upon the literature through better data that includes timing of the trade, size of the trade, and the type of trade (i.e. buy or sell). However, their main focus is not on how insurance impacts liquidity. They do consider insurance by including it as a complexity feature of the bond and looking at how transaction costs change with the complexity of a bond but they do not control for the underlying bond rating.
term index, a long term index, a term that accounts for what type of transaction it was (i.e. buy or sell), and two terms that account for the size of the transaction. They use the coefficients on the size variables and type of transaction to then estimate trading costs. Their result that bond trading costs decrease with credit quality but increases in bond complexity leaves open the question of whether the decrease in trading costs from improved credit quality provided by insurance is enough to o↵set the increase in trading costs from the increased complexity. Another important finding of the paper is that actively traded bonds are not cheaper to trade than infrequently traded bonds, which emphasizes the importance of how one measures liquidity.
Green, Hollifield, and Schrho↵ (2007) is similar to Harris and Piwowar (2006) because they measure transaction costs, but Green, Hollifield, and Schrho↵ (2007) use a di↵erent estimation approach. While Harris and Piwowar (2006) use a time series estimation approach, Green, Hollifield, and Schrho↵ (2007) use a structural model to breakdown the cost of transactions into two parts: the dealer’s market power and the dealer’s cost. This di↵erent approach leads them to similar results as Harris and Piwowar.