Can Consumers Predict Macroeconomic Variables? Gaetan “Guy” Lion August 2010
Introduction <ul><li>The Conference Board asks consumers monthly about their “expectations for 12 months hence” for the following items: </li></ul><ul><li>1) Interest rate (higher/lower); </li></ul><ul><li>2) Stock prices (increase/decrease); and </li></ul><ul><li>3) Inflation (average inflation rate). </li></ul><ul><li>We will investigate how accurate consumers have been in making such directional predictions using the related Conference Board data going back to June 1987. </li></ul>
How to measure the Public’s response: The Relative Value <ul><li>To measure the public’s response, we will use The Conference Board’s “relative value” concept which is calculated as follows: </li></ul><ul><li>% Positive answers/(% Positive + % Negative) </li></ul><ul><li>This entails eliminating neutral answers. </li></ul>
Interest Rate Relative Value Above the 50% line, we interpret that the public expects interest rates to be higher in 12 months. Below it, the public expects interest rates to be lower. Let’s see what interest rates did during this period… The public expected interest rates would increase 12 months hence in 248 months out of 266 months or 93.2% of the time.
Interest Rates Levels We can observe that the majority of the time, rates trended downward.
12 month change in rate level We created a single interest rate proxy by weighting the four rates shown on the previous slide so that the proxy would be more correlated with all four other rates than any other rate vs the other three rates. This was to ensure this interest rate proxy was most representative of the other four rates. Rates increased only 34.2% of the time vs the 93.2% expected by the public.
Picking up the turns This table differentiates between the times when rates continued in the same direction between the previous 6 months and the next 12 months (green area) vs the ones when they reversed course (orange area). It should have been more challenging to forecast correctly when rates reversed course. However, we observe something different. In essence, the public predicted that interest rates would increase almost all the time. By doing so, they seemed to have picked up the turns when rates decreased in the past and increased in the future. However, they were wrong a 100% when the reverse was true.
Contrarian. Taking the opposite of the Public’s prediction The tables above show that one would have been more accurate in predicting interest rates direction by systematically taking the opposite position of the public’s expectations.
Thoughts on rate prediction <ul><li>As shown, the public’s record of predicting the direction of interest rates 12 months hence is poor; </li></ul><ul><li>The public expected rates would increase the vast majority of the time. Instead, they actually decreased the majority of the time; </li></ul><ul><li>Given the current historically low rate levels, if the public sustains its “rate increase” bias it may have more luck in the future (and the contrarian position less so). But, this would be just luck. </li></ul>
Stock Market Relative Value Above the 50% line, we interpret that the public expects stock prices to increase in 12 months. Below it, the public expects stock prices to decrease. Let’s see what the Stock Market did during this period… The public expected stock prices would increase 12 months hence in 215 months out of 267 months or 80.5% of the time.
S&P 500 Level Visually, we can observe that stock market prices as shown by the S&P 500 market level trended upward for a majority of the time.
12 months change in stock price level Stock market prices over the next 12 months rose 74.9% of the time which was not that different than the public’s prediction (80.5%).
Picking up the turns Here we observe an expected pattern. The public did a better job (higher % correct) at predicting stock prices increases or decreases when they continued trending in the same direction vs when they reversed course. Overall, the public did better at picking increases in stock prices vs decreases in stock prices. This makes sense since the market has an upward bias and so does the public.
The Contrarian Position In this case, the Public is far more accurate than the Contrarian prediction.
Thoughts on stock price prediction <ul><li>Here the public did reasonably well as it predicted that stock market prices would increase 80.5% of the time vs the actual 74.9% of the time; </li></ul><ul><li>Given such a record, taking the contrarian position over the reviewed period was doomed to fail; </li></ul><ul><li>Going forward, if we assume that over long periods of time the stock market sustain its upward performance the public is likely to continue its reasonably good performance. </li></ul>
Predicting Inflation This section is different. This is because the Consumer Board does not ask consumers whether inflation is going to be higher or lower. Instead, it is asking the consumers what will be the inflation rate 12 month hence? So, consumers or the public make a specific inflation rate forecast. The metric we use is change in CPI over next 12 months.
A couple of related stats… We used the public’s inflation prediction as the independent variable to predict the actual inflation rate in 12 months. The regression statistics confirm that the public’s predictions are completely different than actuals. The predictions are directionally wrong (negative slope -0.573; instead, it should be closer to + 1.0) and typically overstate inflation (intercept 5.6%; it should be closer to 0.0%).
Concentrating on period since August 1999 Since August 1999, the public has predicted an increase in inflation rate. Meanwhile, overall inflation has trended downward.
Since 1999, a few related stats… When focusing on the period since 1999, everything is now more pronounced. The public’s prediction overstate inflation even more (intercept 7.5%). And, directionally it is now completely wrong (negative slope -1.013). The latter is readily apparent on the scatter plot.
Correlation Matrices As shown, the correlations between the three variables vs the correlations between the three sets of predictions of those same variables are very different.
The Public’s Upward Bias <ul><li>The public has an upward bias when making the mentioned predictions. This may result in descent predictions for interest rates given that we are in a low rate environment. It may also work for stock prices given the market’s upward bias over the long term. For inflation given the public’s upward bias combined with its directional errors, its predictions may continue being off the mark. </li></ul>