The dominant factor in determining interest rates on thirty year mortgages in the interest rate on ten year Treasury bonds.
This isn't too surprising. The average convention thirty year mortgage is actually in place for ten years, and both are among the lowest risk investments in the financial system. Treasury bonds capture both the hypothetical risk free rate of return in terms of real time value of money preferences, and the market's assessment of ten year time period inflation expectations.
The deviations from the rule are mostly at the high end, when Treasury bond rates are around 10% or more, and mortgage interest rates are around 12% or more. This may reflect the unpredictability of inflation and expectations of strategic behavior by mortgage holders when interest rates are temporarily very high.
The Federal Reserve has recently embarked on a more than trillion dollar campaign to buy mortgage backed securities, an unprecedented effort that should lower mortgage interest rates, and indeed it has had that effect in a measurable way. But, does it really make sense to make the kind of massive investment in mortgages that the Fed has made simply to secure the results, which is a roughly 1/2 percentage point reduction in conventional thirty year mortgage interest rates for roughly a year?
The big picture lesson here, in my view, is not that massive efforts by the Fed can produce measurable results. Instead, it is that even Fed efforts on an unprecedented scale that involve a large share of the entire market have only a slight impact on mortgage interest rates. Individual participants in large financial markets, even participants as big as the Fed, have little ability to fight the powerful inherent tendencies of the market. Market trends are very difficult and very expensive to counter through market participation. Market participation is a very inefficient way for government to intervene in the economy.