Not a lot seems to have been written about the effect of the slowdown of the economy of Singapore on the media industry and its allied professions. (Either that or I have been quite late with the news.)
In any case, I have decided to take matters into my own hands and looked at the trends in the media marketplace. What I did is looked at the past four years in terms of advertising spends for those which have been monitored by Nielsen Media Research.
Here is what I saw:
Looking at the above trends, we a generally upward, linear trend in the way advertisers are investing in the media marketplace - though several things caught my attention:
- There appears to be some volatility in the data since 2005. There is some seasonality - particularly in the middle of the year during the summer holidays and towards the last quarter, we see a general uptrend.
On the opposite trend, we see that post-Chinese New Year, we see that investments in the media marketplace declined dramatically. This post-New Year could very well a "recovery phase" after having concentrated moneys in the fourth-quarter and the weeks leading to the festive season.
- As soon as we entered 2009, we began to see a deviation from this 'average' upward trendline.
If you look at the last data point, you will see that there is a significant deviation from the linear expectation for January and February 2009.
This led me to the question: So what is causing these deviations?
The deviations for the first two months of 2009 - and perhaps, even the second half of 2008 - could very well be explained by the onset of the financial crisis and the recession. But then again, the inquisitive - and quantitative-orientated - mind asked: "Can that be quantified? Can the effects of the global and local slowdown be quantified?"
Together with some members of the UM Singapore team, we looked at possible explanations. Using good old Microsoft(R) Excel, we came up with several models incorporating different variables that could explain the variations of the marketplace in the market.
After reviewing more than 10'000 combinations of economic indicators - such as GDP growth, seasonality, consumer price index (which we assumed to be an indication of demand and supply of basic commodities), retail sales and catering indices, and others from SingStat, arrived at a model that was 'good enough' and has explanatory power.
The model above compares the actual (in blue) and the modelled values. The adjusted R-squared - an indication of how much variance could our model explain - amounted to almost 90% (0.889, to be exact). In addition to this high adjusted R-squared, other goodness-of-fit indicators were also significant.
(Visually, it was able to explain the significant decline in January and February 2009 - which is the most problematic data point. There was one outlier that we had to sacrifice: June 2007. It appears that June 2007 saw a significant increase that our model couldn't explain.)
As with all models, we wanted to go beyond pure fits. We wanted explanations. And we broke down the effects of each of the different components of the model.
I cannot divulge the entire model and the equation, but here is a snapshot of the model's breakdowns:
Several things to note from the analysis of the past four-years' advertising media investments in Singapore:
1. There is a general uptrend - as I have noted above. We call this the innate momentum of the marketplace. This innate momentum is significantly strong to drive the media industry upwards in the future. Our tests indicate that even in the rest of 2009, this innate momentum could potentially be the base on which the market will build on. Whilst it is on an uptrend, it explains only about 40% of the total movements in the marketplace's investments in traditional meda.
2. There is seasonality in the marketplace with each month contributing significantly. In fact, if you looked at the last two months in the data set - January-February 2009 - it was the seasonality factor that held the fort together with the innate momentum that we have uncovered.
2a.Special events (which would not be a secret to those with keen eyes) would drive some uplifts - in addition to seasonality.
3. There are two economic factors that we have uncovered that affect the media landscape's investments - one (the green one) accounted for almost 40% as well of the movements in the investments in the past five years. The green economic factor was quite strong that it had 'spillover' effects up to two months.
The other one - in purple - had very limited effects on thye changes in the media industry. If you look closely however, the purple factor could hold the industry on its own. See, for example, August-October 2007: the green economic factor was on the decline, but the purple economic factor was contributing a positive impact on the total media investments.
The next question...: What's going to happen next?
A very good question, really.
The next step is to identify what's going to happen in the future. But remember: our models were based on economic indicators and at least one economic indicator accounted for about 40% of the data's movements. That makes predicting the future of the media industry's size in 2009 and 2010 even more difficult.
But my team and I may just have fuond a way.
That will be for another day.
The Usual Caveats
These are based on my personal estimates and are not necessarily reflective of the company and brands that I work for. The models - whilst acceptable from a statistical and econometric POV - are just models: they cannot explain everything. And as we learned from the financial crisis that is now gripping the world, the past is not necessarily a good predictor of the future and hindsight is always 20/20.
These are what I personally call "descriptive models" - they describe what have happened in the past with the hope that we learn something.
I used Nielsen Media Research data on this - thank you very much. In addition, the econometric modeling was done using a combination of pure manipulation of numbers on Microsoft Excel and a proprietary tool of UM/Initiative.