Blog - Investing Notes
February 12, 2007 - POMO, TOMO and Stock Returns
A reader hypothesizes that the Federal Reserve uses Open Market Operations repurchases to stimulate, or prop up, the stock market. The hypothesis supposes that private parties, such as prime brokers, use the funds released by these repurchases to buy (highly leveraged) stock futures contracts, immediately attracting arbitrageurs who simultaneously short futures and purchase stock indexes. Trend followers then pile on. The Federal Reserve states that these operations regulate "the aggregate level of balances available in the banking system," thereby keeping the effective Federal Funds Rate close to a target level. The operations are predominantly repurchases, whereby the Federal Reserve provides liquidity. Do these Permanent Open Market Operations (POMO) and Temporary Open Market Operations (TOMO) affect the stock market? Are the managers of POMO and TOMO transactions an overachieving "Plunge Protection Team?" Using accepted transactions data for POMO (available since 8/25/05, 51 transactions) and TOMO (available since 7/7/00, 1926 transactions) and contemporaneous daily closing S&P 500 index data, we find that:
We start with TOMO activity because the available historical data sample is much larger and longer in duration. The following chart shows the contemporaneous daily behaviors of TOMO activity and the S&P 500 index. We omit data for the week after 9/11/01 because extremely large TOMO transactions after the terrorist attack (perhaps an unequivocal plunge protection action) dramatically expand the chart scale and obscure any patterns in TOMO activity that might be visually evident. Because TOMO activity occurs almost every day, we insert zeros for the few days with no activity. While TOMO activity mostly rises over time, the daily activity is very noisy. No relationship between aggregate stock prices and TOMO activity is visually evident.
For a more sensitive test, we use a scatter plot.
The following scatter plot depicts the relationship between daily TOMO activity and S&P 500 index returns the same days (TOMO transactions normally settle the same day). Again, we omit TOMO data immediately after 9/11/01 because it generates extreme outliers that make visual inspection of the plot difficult. If TOMO activity noticeably stimulates stock market prices over the same trading day, we expect a significant positive correlation with same-day stock returns, resulting in a distribution sloped from the lower left to the upper right. However, the plot appears to be approximately symmetric around the horizontal axis. In fact, the Pearson correlation between these two series is 0.05, indicating no relationship.
Perhaps there is an effect other than same-day, so we experiment with various lead-lag scenarios.
The next chart shows the Pearson correlations for various lead-lag relationships between daily TOMO activity and daily S&P 500 index returns. If a falling stock market stimulates TOMO activity (i.e., Federal Reserve attempts at stock market manipulation), we expect negative correlations between TOMO activity and near-past stock returns. If TOMO activity boosts stock prices, we expect positive correlations between TOMO activity and near-future stock returns. However, correlations for all lead-lag scenarios from ten days before to ten days after TOMO transactions are near zero, indicating that stock market returns do not influence TOMO activities, and TOMO activities do not affect stock returns.
In case the effects of TOMO activity may emerge only over longer periods, the next chart shows monthly cumulative TOMO transactions (including this time the data around 9/11/01) and monthly S&P 500 index levels. The spike in TOMO activity after the terrorist attack is obvious, but no other relationship between TOMO activity and stock prices is visually evident.
Again, we use a scatter plot for a closer look.
The following scatter plot depicts the relationship between monthly cumulative TOMO activity and same-month stock returns. The Pearson correlation between these two series is 0.01, indicating no relationship over the course of full months. Without the 9/01 outlier, the correlation is 0.16, which is positive in support of the original hypothesis but still unimpressive (R-squared = 0.03) for a sample of only 78 observations.
Next, we examine the smaller POMO activity sample.
The next scatter plot shows the relationship between daily POMO activity and S&P 500 index returns the same days. Because POMO transactions are relatively infrequent in the sample (8/25/05-2/8/07), we do not insert zeros for days when there is no activity. The Pearson correlation between these two series is 0.14, which is positive in support of the original hypothesis but still unimpressive (R-squared = 0.02) for a sample of just 51 observations.
In fact, POMO transactions usually settle the next day, so correlation with next-day stock returns is probably more appropriate. The Pearson correlation between POMO activity and next-day stock returns is -0.05, indicating no relationship.
Might other lead-lag relationships be informative?
The final chart shows the Pearson correlations for various lead-lag relationships between daily POMO activity and daily S&P 500 index returns. The variation in the correlations appears patternless and is probably just noise readily visible in a modest sample of 51 observations.
In summary, analysis of available data does not support the hypothesis that the Federal Reserve attempts to manipulate the stock market via its Open Market Operations.
For examinations of other potential trading indicators, see Blog Synthesis: Some Trading Indicators. See also our blog entry of 12/20/06, concluding that M2 money supply is not a useful indicator for concurrent or intermediate-term future stock market returns.
