- John Hawver

# Curve Ahead

There’s been a lot of noise in the markets lately about the inverted yield curve. The thesis being that an inverted yield curve usually leads a market downturn and/or recession.

Without going into all of the economics of why this is so, it makes some intuitive sense. Why would someone be willing to pay you higher interest in the short-term than in the long-term? Banks usually borrow in the short-term (so they pay that rate) and lend in the long-term (where they earn that rate). If the short-term is above the long-term then it should (and is) bad for banks. (This by the way is the exact thing that caused the savings and loan crisis in the 1980’s.)

Why would a bank do this? Well, aside from the Federal Reserve setting the short-term target rates, the bond market is sending a couple of messages, and I’ll simplify. First, that it is riskier to lend in short-term than in the long-term, there is some expectation of economic troubles in the short term, but the long-term should be ok. Second, in the short-term, more inflation is expected, but in the long-term, less inflation, or that the deflation in the short and medium-terms gets integrated into the longer-term. Neither of those two messages is hopeful for the stock market or economy.

But, oddly enough, the lower long-term rates go, and I’ve mentioned this in the last post, equity valuations rise. So, the message is a bit mixed.

Ok, so, I set out to see if Mr Market was actually being rational. Trusting anecdotes to invest your money probably isn’t the best way to run your portfolio, whatever level investor you are.

To do so, I regressed the 3-month / 10-year curve difference with 1-year forward equity returns, since 1982. This data was easily obtained from FRED and I’ve included the R code below.

What I found wasn’t the stuff to make you very confident. The adjusted R2 of the regression was 1.62%. Basically, only 1.62% of the forward returns can be explained by the level of the 3m-10y curve level. Not stuff you want to bet the farm on.

But, like Twain said, statistics can always deceive, so let’s dig a bit deeper. The slope of the regression line makes sense, it is positive, so as the curve rises so do forward equity returns. That’s good. More impressively, if we subset the data, and only look at the times when the curve is inverted, we find that 61% of the time the 1-year forward returns are negative. That’s better than a coin flip and starting to be meaningful.

All in, I’d say the anecdote has some merit. Should the level of the curve be the ONLY indicator used to forecast forward returns, absolutely not.

When you combine this indicator with other factors in the market such as the high levels of valuation and the jobs picture, and maybe this is my own confirmation bias coming through, I think it does merit some caution going forward. But certainly not the full stop the media would have you believe.

No single indicator can predict the market no matter how much we’d all like to try.

# Get data

idx <- getSymbols('WILL5000IND', src = 'FRED', auto.assign = F)

crv <- getSymbols('T10Y3M', src = 'FRED', auto.assign = F)

# merge and clean data

all_data <- na.omit(merge(idx, crv))

names(all_data) <- c('idx', 'crv')

# add returns and lead

all_data$idx_lead <- xts(fLead(coredata(all_data$idx), 252), index(all_data))

all_data$rtns_lead <- (all_data$idx_lead - all_data$idx) / (all_data$idx)

# regress

Y <- all_data$rtns_lead

X <- all_data$crv

mod <- lm(Y ~ X)

summary(mod)

# plot

ggplotRegression <- function(fit) {

ggplot(fit$model, aes_string(x = names(fit$model)[2], y = names(fit$model)[1])) +

geom_point() +

stat_smooth(method = "lm", col = "blue") +

labs(title = paste("Adj R2 = ",signif(summary(fit)$adj.r.squared, 5),

"Intercept =",signif(fit$coef[[1]],5 ),

" Slope =",signif(fit$coef[[2]], 5),

" P =",signif(summary(fit)$coef[2,4], 5)))

}

ggplotRegression(mod)