Twenty years ago, having just arrived in Dallas, I wrote a column predicting 1985 would be a very good year for stocks.

My analysis was basic. Interest rates were likely to continue falling. As a consequence, investors would be willing to pay more for a dollar of earnings. That meant price-to-earnings multiples could rise even if earnings didn't rise.   Investors would enjoy price appreciation through P/E multiple expansion.

And that's what happened.

The total return on the S&P 500 stocks in 1985 was an out-sized 32.2 percent. It was followed by an 18.5 percent total return in 1986.

While many investors don't remember what 1985 felt like, stocks had tough competition from bonds and certificates of deposit. Although interest rates had peaked several years earlier, short-term U.S. Treasury obligations were yielding 9 percent. Quality long-term bonds were yielding about 12 percent. Since the historical average return on stocks was 10 to 11 percent, many investors preferred the safety of bonds and CDs.

Few wanted the uncertainty of stock returns. While equity mutual fund assets increased by 38.7 percent during 1985, indicating a modest net flow of new money on top of the 32.2 percent return, bond fund assets nearly doubled. Stocks were cheap--- but most people were afraid to buy.

Today the situation is reversed.

Bond yields are low. Short-term Treasury obligations earn less than 2.5 percent. Long term Treasuries earn less than 5 percent. According to the Leuthold Group in Minneapolis, these yields rank in the bottom 10 and 30 percent of all quarterly time periods since 1957, respectively.

Today, if interest rates rise, stocks will face increasing competition from bonds and certificates of deposit. If the long decline in interest rates--- one of the most powerful drivers for our economy and the stock market--- is over, stock P/E ratios are likely to shrink.

The higher bond yields go, the less investors will be willing to pay for a dollar of earnings. We may be at the beginning of a long P/E multiple decline. Some, but not all, of the decline can be offset by rising earnings. The important thing to remember, however, is that earnings multiples are likely to shrink, not expand.

You can understand the impact of bond competition by using one of the many models for valuing common stocks. The Goldman Sachs model, used in my column 20 years ago, assumes that investors demand a return on their stock investments that is equal to the yield on a bond plus a risk premium. If bonds are yielding 5 percent today and the risk premium is 5 percentage points, a total return of 10 percent, stocks would have to be priced to provide that return from growth in book value and the dividend yield.

In fact, the algebra nearly blows up in the current market. With a return on shareholder equity of nearly 13 percent for the S&P 500 index companies and 70 percent of earnings retained, stocks need a dividend yield of only 0.9 percent to provide a 10 percent annual return. Since the current market yields about 1.7 percent, stocks appear to be undervalued by this measure.

Needless to say, other models and other indicators don't agree.

What's important, however, is to see the impact of rising interest rates. If interest rates are 5 percent, the model says we can pay 33 times earnings for the S&P 500. But if rates rise to 6 percent, the same model says we can only pay 16 times earnings. It indicates only 10 times earnings for 7 percent bond yields and 8 times earnings for 8 percent bond yields--- about where stocks were priced at the 1982 market low.

In fact, Standard & Poor's estimates 2005 earnings for the S&P 500 at $74, which gives a future p/e of 16, compared to the current multiple of 18.

What does that mean for you and me as investors?

Two things. First, valuations are likely to be reasonable in 2005 even if long-term interest rates climb to 6 percent. If interest rates remain the same, 2005 could be another good year for the U.S. stock market.

Second, the caution light is on. We need to watch interest rates very, very carefully.