The pattern and the puzzle
Look at a set of areas that received renewal investment and a set that did not, and a common finding is that the renewed areas often show more economic activity afterward: higher property values, more businesses, more jobs. It is natural to conclude that renewal caused the growth. But a correlation, meaning that two things tend to occur together, does not by itself show that one causes the other.
This matters for decisions. If renewal truly causes growth, spending on it is an investment. If it merely accompanies growth that would have happened anyway, the spending may add little.
Four reasons the numbers can mislead
1. Selection effects
Renewal projects are not scattered at random. Governments and developers choose where to invest, and their choices are shaped by prospects. An area chosen because it is near a growing downtown, has good transport or has a motivated community may do well with or without the project. In that case the project is following growth, not causing it.
Selection can also run the other way. Areas chosen because they are in severe decline may keep struggling despite investment, making renewal look ineffective when it may have prevented a worse outcome.
2. Confounders
A confounder is a third factor that influences both renewal and growth. A strong regional economy is the classic example: it raises tax revenue and developer interest, making renewal more likely, and it also lifts local businesses and property values on its own. Without accounting for the regional economy, the correlation between renewal and growth is partly just a shared response to it. Other possible confounders include population trends, interest rates, national policy and local leadership.
3. Reverse causation
Growth can cause renewal, not only the reverse. Where demand is rising, developers want to build and cities have the revenue to improve streets and buildings. Areas that are already growing attract renewal. Taken at face value, the data would credit renewal for what was in fact a response to growth.
4. Measurement and averages
What is measured shapes what is found. Property values may rise in a renewed area because lower-income households left, a change that raises averages without improving anyone’s circumstances. Counting the area’s jobs, not residents’ jobs, can overstate local gains. Different time windows can tell different stories. See the discussion of regeneration metrics for how particular indicators can mislead.
A small hypothetical
This example is invented for illustration. Suppose a city has ten neighborhoods, and renewal is carried out in the four that the city judges most promising, because they are close to jobs and transit. After five years, the four renewed neighborhoods show stronger growth in property values than the six others.
Does this mean renewal worked? Not necessarily. The four were chosen because they were promising. If they would have grown faster anyway, some or all of the gap reflects their advantages, not the project. To reach any conclusion, one would want to know how the four compared with the six before the renewal began, and whether similar neighborhoods that were not renewed grew at similar rates. If the four were already pulling ahead before the project, the gap afterward is weak evidence of an effect.
Research designs that try to do better
Researchers use several approaches to get closer to cause and effect. None is perfect.
- Before-and-after with comparison groups. Compare changes over time in renewed areas with changes in similar areas that were not renewed. This helps if the areas were on similar paths beforehand.
- Matching. Pair each renewed area with a non-renewed one that is similar on many characteristics.
- Natural experiments. Look for cases where investment was decided in a way that is unrelated to local prospects, such as a funding lottery or a cutoff rule. These are rare but informative.
- Tracking people, not just places. Follow the same households and firms over time to separate real improvements from changes in who lives there.
- Looking at timing. If growth began before the project, that weakens a causal story.
Even with these methods, results are estimates with uncertainty, and findings in one place and period may not carry over elsewhere. A pilot project can sometimes be set up in a way that supports clearer comparison.
Why it matters for the public
Understanding the limits of correlation helps in evaluating claims.
- A report that an area boomed after renewal says little about cause by itself.
- A claim that renewal failed because an area stayed poor also needs a comparison: what would have happened otherwise?
- Differences in outcomes among places may reflect where projects were put, not only how well they were run.
- Policy based on weak causal evidence can waste money or, conversely, discard something useful.
It also does not follow that renewal does nothing. The honest position is that the research on renewal’s effects is mixed, with some designs finding gains in some settings and others finding little. Effects can be real but small, large for some groups and absent for others, or sensitive to context.
Questions to ask of any claim
- How were the renewed areas chosen?
- What was happening in those areas before the project?
- What is the comparison, and how similar is it?
- Were other changes happening at the same time?
- Does the measure reflect people or places?
- How long were the effects observed?
Summary
Renewal and growth often appear together because renewal tends to go where growth is likely, growth can attract renewal, and common forces drive both. Distinguishing real effects from coincidence requires comparison groups, attention to timing and care about what is counted. Without those, a correlation is a prompt for questions rather than an answer.