Macro Roundup

"...A little less than40 per centof the mortgage debt owed by American households was held directly on the balance sheet of “US-chartered depository institutions” in the years before the crisis. That share has since dropped to just over 30 per cent:..."

Office212.404.4646

Mobile908.303.9747

The more useful takeaway is that the trend highlighted by the Boston Fed economists, combined with anecdotes about changes in underwriting standards and the growth of compliance staff, suggests banks may have learned something useful from the crisis. They are willing to spend more today to ensure they avoid legal liabilities for negligence and “misrepresentation” in the future. That would be welcome news, even if it means consumers are “paying” more for credit up front than they did in the past.
--

The Boston Fed authors think it might help to have morefloating-rate mortgages, which would obviate the need to create an entire new loan whenever a borrower wants to take advantage of a drop in interest rates. From their point of view, more floating-rate debt would improve the transmission of monetary stimulus in a downturn by putting more money in the hands of mortgage borrowers rather than the banks. It doesn’t seem particularly helpful to suggest such a thing now, however, when the Fed seems determined to push interest ratesup.

This relationship between mortgage rates and refinancing createsnegative convexitythat investors tend to hedge by selling Treasury bonds. Of course, most of the post-crisis agency MBShas been bought by the Fed, which doesn’t have to worry about marking its holdings to market, which might help dampen some of the traditional volatility caused by these hedges.

These two factors are partly related, since the rate of mortgage applications is affected by announcements of Fed stimulus. After all, falling mortgage rates makes refinancing a lot more attractive. In the post-crisis period, the rate of mortgage applications has been strongly related to changes in the spread between what existing borrowers pay on their mortgages and the 10-year Treasury yield:

Their charts below show how the “price of intermediation” has tended to move in line with these constraints. On the left, you can see how banks tend to respond to an uptick of mortgage applications with slower processing times. The typical processing time (red line) seems closely related to the “price of intermediation” (gray line). On the right, you can see how wages in the sector rose rapidly during the period that borrowers were getting charged more:

Second, banks have limited capacity to properly underwrite new mortgages. Sudden increases in loan applications can’t easily be accommodated without hiring more people or paying existing staff more. Borrowers end up being the ones who cover these costs.

When the third round of large-scale asset purchases was announced inSeptember 2012, only about half of the large drop in MBS yields got passed along to borrowers. By contrast, the full brunt of the “taper tantrum” was felt by prospective mortgagers. (Some might argue this was helpful insofar as it helped recapitalise banks still wounded from the crisis.)

First, the Fed’s aggressive purchases of agency MBScrushed credit spreads so muchthat banks could lower interest rates to prospective borrowers while also pocketing some of the collapse in their funding costs for themselves.

The Boston Fed economists attribute this asymmetry to two main factors.

A one dollar decrease in the MBS price translates into a one dollar reduction in the rebate, but a one dollar increase only leads to an 80 cent rebate increase.

In general, changes in MBS prices flow through quickly to changes in what banks lend, although this is more obvious on the way down than on the way up:

The difference between the interest rates on mortgages and the yield on the MBS, adjusted for the g-fees, is therefore a pretty good proxy for the cost of what banks actually do. More precisely, the banks get paid the difference between what they raise when they sell their mortgages to the GSEs and what they actually lend to borrowers (the rebate). This is what the Boston Fed researchers are studying.

(Fannie and Freddie got into trouble during the crisis because they had charged insufficient g-fees during the bubble and were insufficiently capitalised to compensate for that mistake. Making matters worse, the banks had sold dubious loans to the GSEs, which led the Federal Housing Finance Agency to repeatedly sue the banks for “misrepresentations” after the bubble had burst.)

Investors in typical agency MBS of “conforming” mortgages don’t have to deal with the risk that borrowers will default. That risk is instead borne by taxpayers and ostensibly covered by the guarantee fees paid by borrowers to the banks and then to the GSEs.

The pre-bubble and the post-crisis norm is for most securitised mortgages to be issued by the government-sponsored enterprises of Fannie Mae and Freddie Mac. The yield on these agency MBS is more or less the cost the banks pay to fund their mortgage lending. Unsurprisingly, this is lower than the interest rates paid by borrowers on the mortgages bundled into a given bond.

That worked out poorly for those investors and the market has essentially disappeared since the crisis. Almost the entire net decline in the stock of mortgage debt since the peak in 2008 has come from defaults on mortgages packaged into private-label MBS:

During the go-go years, abouthalf the growth in mortgage borrowingwas funded by the growth of “private-label” securitisation, which was when banks packaged loans and sold them directly to investors:

A little less than40 per centof the mortgage debt owed by American households was held directly on the balance sheet of “US-chartered depository institutions” in the years before the crisis. That share has since dropped to just over 30 per cent:

The Boston Fed research focuses on the fact that most American mortgages are sold to investors after being bundled into securities, rather than held by banks on their balance sheets.

These were choices. In the go-go years, there didn’t seem to be much point risking market share to competitors who didn’t spend as much on their back office. After all, if investors didn’t care about underwriting, why should the banks? And it wouldn’t matter if a borrower’s income were insufficient to meet his mortgage payments, since rising house prices would “guarantee” profits for mortgage lenders even if they had to foreclose and sell the collateral at a discount. These were bad assumptions, but they were the consensus.

Banks failed to do these jobs properly both before and during the crisis. Their underwriting was poor, which is one reason household debt ballooned, particularly among people who weren’t able to repay it. And when the bubble began to deflate, banks lacked the staff to ensure they followed the rule of law, modify mortgages, and do what was best for creditors. (The government’s apparent lack of interest also didn’t create much incentive for the banks to change their approach.)

That sounds bad, but these extra costs may be there for good reason. Banks play an important role in American housing finance even though most of the credit is provided by others. Banks are supposed to vet potential borrowers, set appropriate terms for loans, collect payments, and foreclose in the event of default. (Unfortunately they also have a track record offoreclosing on people who didn’t default, and sometimes even on people who owned their homes outright…)

Researchfrom economists at the Federal Reserve Bank of Boston finds this spread has steadily increased since the crisis, with American borrowers effectively paying banks an extra $140 billion in 2009-2014 above what you might have expected given historical patterns. As they put it, “over extended intervals during our sample period, mortgage rates would have been 30-40 basis points lower under the counterfactual”.

It’s a quaint misconception that banks fund mortgages. They do, however, affect thepriceof mortgages by charging a spread over their cost of funding.

Matthew Klein, "Post-crisis mortgages cost more but maybe that’s a good thing,"FT Alphaville, April 7, 2017, https://ftalphaville.ft.com/2017/04/07/2187034/post-crisis-mortgages-cost-more-but-maybe-thats-a-good-thing/ Post-crisis mortgages cost more but maybe that’s a good thing

Change In Income Distrbution Driven By > 99.99% 12,000 Household. The Share Of Incomes Above The 99% ~ $372,000 in 2012 But < 99.99% ~ $7.2 Million in 2012) Has Barely Changed In Last 20 Years

Fatih Guvenen and Greg Kaplan, "Top Income Inequality in the 21st Century: Some Cautionary Notes," April 2, 2017, https://gregkaplan.uchicago.edu/sites/gregkaplan.uchicago.edu/files/uploads/top_income_inequality_web_April2_2017.pdf

"...Since 2000, different measures of top income inequality have exhibited very different trends. Top income shares based on measures of total income show a continued rise, whereas top income shares based on wage and salary income show no increase in inequality post-2000. The most important difference between these two measures of income is the income that accrues to S-corporations.... But interpreting trends in the S-corporation component is extremely difficult. Feenberg and Poterba (1993), Gordon and Slemrod (2002), and Cooper et al. (2016) warn that much of the recent increase in S-corporation income is income that previously accrued to C-corporations. Such income is not “new” income earned by top earners but is simply income that was previously labeled as corporate income rather than household income....Moreover, the majority of the recent increase in top income shares is due to an increase in average earnings at very high income levels, much higher than that assumed in typical discussions of top income inequality. Once incomes above the 99.99th percentile are excluded (around $7 million in 2012), we see that little continued growth in top income shares has taken place in the last 20 years. Put simply, so far in the 21st century, all the action in top income shares has been S-corporation income at very, very high income levels..."