James Anderson leads the Government Innovation Program at Bloomberg Philanthropies, the umbrella for the charitable giving of billionaire and former three-term New York City Mayor Michael Bloomberg.
He was Mayor Bloomberg’s communications director, leading on the design of NYC Service and on public engagement for a number of Bloomberg reforms.
James has paid more attention than almost anyone to how cities work, and how they learn from each other. But is the Bloomberg model for making cities better “technocratic”? What can it do, and what can’t it do? And should mayors be “innovative”? Or are the best practices, at the end of the day, pretty straightforward? We get into these questions and more.
Harry Fletcher-Wood, Shadrach Strehle, and Jasper Placio helped produce this episode. This transcript is lightly edited for clarity.
For a printable PDF of this interview, click here:
How do you describe the work you do?
I run a portfolio of capacity-building programs called the Government Innovation Programs. They equip mayors and their teams with the capabilities they need to solve problems better.
We run leadership programs that train everyone — the mayor, the chief of staff, the budget directors.
We focus on great organizations, equipping cities with better data practices, innovation practices, and teams.
We help cities generate and test more ambitious urban solutions. If they work, we help them spread to cities around the world.
We’re building a better field of organizations that focus on strengthening the capacity of local government, and building a tremendous amount of research on what works in cities.
Across all of that, we focus on making sure that mayors are constantly learning from each other and building on one another’s successes.
Our program started in 2010, with the first five grants. Mayor Emanuel in Chicago and Mayor Landrieu in New Orleans got innovation teams. Today, more than 900 cities are getting deep technical assistance grants and leadership development from us. Thousands more cities are benefiting from our education programming.
Almost every issue that affects people in their daily lives — housing, safety, climate resilience, economic mobility — is experienced locally; it’s implemented through municipalities. City halls are the first line of response and the last mile of delivery.
But municipalities are stuck using operating models that were developed in a very different era and have been underinvested in. If we want national policy to land in the real world, we have to think about the capacity of municipalities to solve problems in a very different way. That’s what we are focused on at Bloomberg Philanthropies.
In the American system, mayors are the elected leaders with the most power over the domains they’re elected to govern. A representative in the House has about 1/400th the decision-making power over a given bill, whereas a mayor often has a broad range of tools.
What do individual mayors have the power to do?
When you’re working in local government, state capacity variability is your best friend. Oftentimes two cities sitting across county lines have a different authorising and operating environment, because there are different labor regimes, tax bases, powers and authorities. There’s extraordinary variability.
You mentioned the old operating model that many municipalities have. Can you explain that model?
I’ll answer it on two levels, and one level will be super familiar to your audience. Local governments tend to be highly siloed. They operate vertically and not horizontally. This is a leftover of the Fordist organizational model that defined our public services in the last century.
Perhaps a more useful way to think about it is that local governments in the US are still construed, both by federal funders and in public discussion, as organizations that are responsible for delivering an on-time, on-budget service — not as problem-solving entities being asked by their constituents to solve emerging issues that often take them away from service delivery.
The vast majority of mayors’ budgets are fixed — pensions, labor costs, fixed services. The smallest parts of their budget fund governance improvement, strategic thinking, innovation, IT upgrades. Those are the first areas cut in a downturn.
When you say the Fordist model was adopted in the public sector, can you give me more colour? Everyone fulfills a specific role in a manufacturing process? How did we set up cities to be like the Model T production line?
We’re talking about standardized services for a mass public, clear division of labor, and rules-driven bureaucracy. Success is measured by throughput, not necessarily by outcomes. We live in a world where mayors are dealing with transformation — pandemics, mass migration, the affordability crisis — and they’re being asked by their residents to respond in real time. That requires people to work across sectors, across silos, and be more iterative. We have to test, learn, and adapt. This is not the curriculum for the standard Master’s of Public Policy (MPP) or Master’s of Public Administration (MPA) class.
Much of what the public sector has to do today is help communities navigate change, and thrive in spite of that change. You can’t look to a single federal funding stream to enable you to do that. If your budgets are tied to those vertical funding streams — 50-75% of all municipal budgets are fixed or near-fixed costs — the amount you have to develop internal capacity to respond to change is out of that smaller slice.
We recognized that there was a distinct role philanthropy could play to help show what’s possible, set standards, and create a different North Star around what it means to run an organization that is iterative and adaptive. It has to deliver services effectively. We all know a lot of cities don’t get that right. But it also has to be able to shift resources and change direction in response to the change that a community is experiencing. What does it mean for me as a mayor? What do I need to have on my team in order to navigate that change? That’s the through-line of our programming.
As mayor, your old boss Michael Bloomberg was most often described as “technocratic.” It’s funny that the same word is used to describe the old Fordist model, which he rather disdained.
It’s a bit of a lazy description of Mike. He was deeply values-driven. Sometimes “technocratic” was applied because people didn’t know what to do with a mayor who was not an ideological purist. He was deeply interested in finding the best solution to each problem on its own terms. [Statecraft spoke about Bloomberg’s mayoralty recently with another former member of his team, Maria Torres-Springer.] That gets reduced to the word “technocratic,” but to me, there was always a very clear sense of values driving the work that the Bloomberg administration got behind.
That makes sense to me. I don’t share Mayor Bloomberg’s politics, but in our work at IFP, we also get labeled technocratic. I’d have a similar answer to you: we have a specific set of things we care about. We’re very public: “These are our values.” Then we try to be very focused on what success would look like in each of those issue areas and how we can do it. I’m open to being convinced that we should use a different tool for each of those projects. If that’s technocratic, so be it.
One of the key lessons I learned from Bloomberg was, “Understand the problem deeply. Circle that problem with data, and with different perspectives, before you generate a solution.” I was recently speaking with Tim Kelly, the Mayor of Chattanooga. He came to the Bloomberg Harvard City Leadership Initiative. This is one of our flagship programs: it trains 40 mayors every single year, emphasizing the role of managing well: how do you squeeze more efficiency and effectiveness out of the limited resources that you’ve got?
Kelly majored in innovation in the Bloomberg Harvard program. He sent a team; they got coached on evidence-based strategies to understand a problem. He told me, “I’m a results mayor, I’m beating my chest, asking for solutions. That process taught me to use data — the perspectives and experiences of other cities — to understand a problem in a way that I never would’ve understood it before.” Once they did that, they realized that youth crime in their city was overwhelmingly about mental health. That unlocked a whole set of interventions they otherwise wouldn’t have zeroed in on. A while ago, they celebrated a year of zero murders in one of the highest-crime neighbourhoods in their city as a result of those interventions.
When people talk about “evidence-based policy-making,” it’s hard for me to find anyone who will stand up and say, “We’re against it.” So what are you teaching people to do that wouldn’t happen without you?
I didn’t mention evidence-based policymaking. I mentioned evidence-based practices.
Let me answer your question in a couple of different ways. Number one, we’re training mayors to be great managers and training their organizations to be learning organizations. A decade back, that was not in the DNA, was not the way that we thought about mayors, and there was a pressure on city leaders to know all the answers.
Number two, we’ve made huge gains in helping US cities move up the ladder of sophistication in their use of data. One of our programs is called What Works Cities. It’s a certification program. When we started this work 15 years ago, cities around the country said to us, “I don’t know what good data looks like.” Over the last decade, we’ve moved hundreds of cities up that ladder of sophisticated data use. Cities around the US are probably in the best position of many public institutions to embrace the AI era, because they know their data is clean, well-structured, and timely.
Third, when we started there were only a handful of chief innovation officers in the world’s public sector. Everyone you talked to had a different definition of innovation. They were throwing spaghetti at the wall. Fifteen years later, a field has emerged where people are using the same language. They understand what tools help them understand problems better and produce more creative ideas, test those ideas, and scale up the things that work. Those are the evidence-based practices that we promote, because they work in the private and public sector.
Making those practices the norm has taken innovation offices out of the periphery and brought them into the center of city halls. Mayors are using them to advance their big priorities. I’ll give you a really good example. Brandon Scott, the Mayor of Baltimore, inherited an Innovation Team we were funding under the previous mayor. He was in the heat of the pandemic when he took office. Cities around the country were hiring large management consultancies to stand up contact-tracing firms.
Mayor Scott had a deep belief that because Baltimore had just experienced a crisis — the prior mayor had been indicted and left office — locals wouldn’t trust outsiders. They wouldn’t open their doors and provide that information. He took that innovation team and said, “Let’s create a home-grown version of a contact-tracing firm.” They were the only major municipality in the country that did it, and he outperformed his peers on every single indicator that mattered: vaccination rates and more.
Now he’s using his Innovation Team to address another top political priority, which is police recruitment and retention. [The Innovation Team’s focus on this began under the previous mayor.] This year, Baltimore has brought on more than they’ve lost — they have a net increase in police officers for the first year in a decade. That comes back to this mayor using these evidence-based tools around, “How do we understand what’s causing our attrition and retention problem, and where do we find the points of leverage to fix them?”
I still have this lingering question about data. Let’s say I’m the mayor of Baltimore, 30 years ago. I assume the city gathers all kinds of fine-grained data. Can you explain the weakness of those data systems originally?
Cities are incredible generators of administrative data. That data is stored and structured in highly variable ways. Sometimes it takes the form of handwritten forms put into a filing cabinet. Other times it’s an Excel spreadsheet or in a CRM. We’ve tried to help cities structure that data in optimal ways, make sure the governance is excellent, then being able to use that data in ways that inform executive decision making. I hate to say it, but I still hear so many new mayors say, “I’m coming into office and I’m flying blind — not seeing the data I need to feel comfortable with the decisions I’m being asked to make.”
What kinds of decisions are they thinking about?
All types of policy and programmatic decisions. “Mayor, we want to shift our sanitation service from Monday, Wednesday, and Thursday to Tuesday, Friday, and Sunday, and the reason for that is this.” The mayor says, “Show me the data.” They say, “This is what we’re hearing from the ground.” Those types of issues.
Part of our management training for the mayors is focused on:
What are your expectations for how you want the organization to use data?
What are the kinds of questions you can ask that get the organization to bring forward more data?
How do you have some patience with the fact that the data at the beginning is not going to be the best, but it’ll improve over time and with ongoing use?
What kind of people do you need to have in place to make sure that yours is an organization that’s well positioned to use data in an impactful way?
It’s everything: people, practices, routines, performance management, governance, how data is being used with things like procurement — to issues like public engagement and communication.
Let’s say I am that mayor and I feel I’m flying blind on when the sanitation services should run. The only data I’m getting is my Sanitation Department saying, “This is what we’re hearing we need to do.” What should I start implementing?
A long-standing and extremely impactful program is our What Works Cities certification. Any US city of 30,000 or more is eligible to participate. First, we help the cities do an assessment: “Let us understand your existing data practices.” We then work closely with the mayor and the team on a set of priority interventions to improve those practices one step at a time.
I’ll never forget — the first mayor that we worked with was Tony Yarber, then-mayor of Jackson, Mississippi. He came into the What Works Cities program and said, “I don’t have a data person, but my comms guy is going to be our data point person.” We taught that guy the basics around how to help this mayor structure data so that he could ask it questions, and get answers that helped to inform his decision making. Today, the data practices are way advanced. 54% of US cities of 100,000 or more are moving up that ladder of sophistication towards certification in the What Works Cities program. There’s a real movement of data people in cities around the country. People understand today that data is important — you cannot do good AI if you don’t have good data governance in place. There’s extreme interest in getting better.
How do you develop better sources of data without driving the people who are sourcing it crazy? Police officers complain about how much paperwork they have to fill out, but you’d like to get more data from officers.
In some ways it’s about structuring and organising data that these institutions already have, but typically aren’t using well. Almost all of the programs that we operate or fund are focused on priority problems for the mayor. In Baltimore it might be public safety; in San Diego it might be housing. The data work follows a basic rhythm, which is:
What’s the problem you’re trying to solve?
Do you have the data that you need to understand that problem?
If not, what are the gaps?
What is that data telling you?
How often do you need to look at that data to understand where you’re standing?
What are your data flows?
How can you make that predictable?
Let’s go back to Baltimore and the violence-prevention work that this mayor has done. Baltimore has seen a fantastic drop in violence post-COVID. Tell me about the work you did with the mayor.
All credit goes to Mayor Scott and his incredible team. This week they announced homicide levels at 50-year lows, and the data point beneath that is probably even more interesting. Violent crime and murders are coming down in cities across the country, but Baltimore’s driving it down fastest.
Mayor Scott has done a couple of things that are textbook examples of how a mayor solves complex problems effectively. The first is, he made very clear that this is his number-one priority. He put himself at the center of decision making and public communication. He set an ambitious goal. His team was nervous when he came out and said, “We’re going to have a 15% year-on-year reduction in murder.” But he knew that if he didn’t challenge the ambition of the system, the system would do less than it might otherwise.
Then he’s put data-driven decision making at the center of everything he’s done. We helped him stand up a world-class data capability and hire a great chief data officer. Mayor Scott can wake up, pull up his dashboard, and understand where things stand. Then he calls his police chief and they talk about real facts in real time and what the next strategies are going to be. Declaring a clear goal, making it his organizational priority and leaving no room for doubt on that, and continuously using data to inform the strategies that he’s delivering has been the path to great progress.
How innovative do you think that story is? The counter-argument would be: Yes, Baltimore has been a fantastic success story for the reasons you mentioned, including a careful data approach and a mayor who is willing to stake political credibility on it. But isn’t that just like other successful attempts to drive down violence in major cities? The Boston success story from the 90s had many of the same elements. CompStat in New York, which we discussed with Peter Moskos on Statecraft, was this model: we’re going to organize the data, be held politically accountable, and hold people down the chain accountable for knowing exactly what’s going on, and having a plan.
I don’t mean this to denigrate Baltimore’s achievements. But is this an example of applying a best practices playbook, or is there something new here?
He’s absolutely embraced evidence-based strategies; group violence reduction has been at the center of all of the success. Baltimore’s starting point was a tough spot. They were suffering from one of the worst homicide rates in the country. This is a mayor that has figured out how to produce the fastest-declining murder rate in the country. It is a come-from-behind success story, and that speaks to the mayor’s management approach and how he has personally thought very strategically about how he shows up as manager-in-chief of a system.
You’re absolutely right, he’d be foolish not to avail himself of strategies that have worked in other places. His starting context was very different than most of those places, and his level of execution success has been top-notch. He’s now taking those capabilities and applying it to housing vacancy, which is the second-biggest problem in Baltimore.
What I’m after here is not so much, “How much credit can we give the mayor for this work?” But at a more philosophical level, how important is innovation? There’s a huge amount of evidence on how to drive crime down, and several American cities have successfully done that. My impression is that that’s often true for other key things we expect cities to do — there is a playbook.
I tend to define government innovation pretty broadly. It’s about making one plus one equal three or more. Sometimes that takes the form of radical, system-shocking ideas. The work we’re seeing around the care economy — cities are deploying a full-throated municipal response to care, which has been the under-attended-to underbelly of our economy forever — is a great example. But a lot of it is also process and partnership reimagination.
In Baltimore, there was a persistent challenge: the squeegee boys — young men who didn’t have income and were out on the streets offering to squeegee windows for pay. This was something that the mayor was hearing a lot about from the business community. He shook the pockets of everyone at the table: “What assets do you have to put towards this solution? How can we coordinate those interventions so that they feel meaningful to these young men, and we can create an inducement to get them off the streets?” That challenge is no more in Baltimore. That’s a process innovation. It’s not a radical paradigm shift, but we think of government innovation in that very broad way.
There’s room for both of your points in that definition. Sometimes it’s stuff no city has ever done before. We have the Mayors Challenge: we’re about to name the 25 best breakthrough ideas that cities from around the globe surfaced. There’s some thrilling stuff there. But most of the day-to-day innovation we see in city halls is business-model, customer-service, and efficiency innovation. Those are critically important, and they too require capacity. People need to understand how to squeeze more out of these systems. There’s a science to it, and that’s the science that we help cities connect to.
Let me ask you about the role of external consultants in how cities are governed. I talked recently with the head of the federal Office of Personnel Management, Scott Kupor. He had good things to say about similar contracts at the federal level, but he feels much of that work could be done better and cheaper internally.
Compare the state of play at the municipal level. Is there a management consultant bloat?
There’s certainly an over-dependency. Municipalities are less able to afford expensive management consultants than their state and federal counterparts. But I am in total agreement. Innovation used to be contracted out to the McKinseys and the IDEOs; data work used to be externalized. Municipalities need to have internal capacity. They have to own their own data, and the freedom of technical expertise on the inside, so that they can be adaptive and iterative as new needs emerge.
The big consultants benefit from these contracts. But are there also forces in city halls that are opposed to trying to bring that capacity in-house?
The major pushback is fiscal. The vast majority of city budgets are dedicated to relatively fixed costs. The amount of discretionary capacity that most mayors have is small. This is the funding that goes to IT modernisation and beefing up your data — your in-house capabilities. These are difficult fiscal choices. That’s where the reluctance is. We try to make the case that these investments pay dividends over time, and there’s growing evidence that they do.
Are there cases where you have helped city halls stand up these internal capacities, and then a fiscal crunch comes and the capacities are on the chopping block? How do you try and stop those things from being zeroed out in crunch time?
We’ve been doing this work long enough that a number of fiscal cycles have occurred. There are times when cities are reducing any and every cost. What’s amazing is that 95% of the innovation grants that we’ve made, once they have stopped, city halls have maintained those lines on city dollars. A growing number of them — a majority now — are lasting across administrative turnover. When people see what this data capacity can do — what it means to have a dedicated problem-solving team that helps a mayor put points on the board that their residents recognize — they work hard to keep it going. Not all the time, but certainly more and more of the time.
When Mayor Bloomberg came into office in New York City, he set up City Hall on an open floor-plan model — that was a big deal at the time. That’s been retained through multiple mayors.
Besides building better data streams and being more evidence-based in their practice, what other things do you think of as best practice — things you’d encourage any mayor to pursue?
Allowing people to work across silos is central to our capacity-building programs. We just exported our mayoral leadership program, which has been operating at Harvard, and created a sibling program over in Europe. I just got a note from the mayor of Izmir, Turkey, who set up a bullpen, as he was so inspired by the way that worked in Bloomberg’s City Hall.
I think public engagement — we are in a moment where city halls feel that the lack of institutional trust is a real problem. How do you show the work? What does operational transparency look like? How do you let your residents understand where things are and what the black hole of local government is doing? That is an important strategy to rebuild trust in government.
You run a New Mayors Program at Harvard, with the US Conference of Mayors. What are the most common mistakes that you see new mayors making?
If Mike Bloomberg were here, he would tell you that the thing that matters more than anything is getting the team right at the outset. Do you have people who are not sycophantish and are willing to tell you hard truths?” If you don’t have that, you’re in real trouble, because pretty soon the waters get deep and you’re swimming for your life.
Are mayors too often inclined to pull together yes-men and women around them because it’s the natural inclination?
Human beings are. We’ve learned that lesson very well in today’s social media world.
I know there’s a lot of collaboration between mayors. There are best practices that a mayor doesn’t have to reinvent — the town a few miles away may be doing an excellent job. But I’m curious about competition among mayors. A few months ago I had a conversation with the former mayor of New Orleans, Mitch Landrieu. As a mayor, he was motivated by the idea of competition: “What are they doing, and can we beat them?”
Do you see inter-mayoral competition often?
This is an amazing force for public improvement. Mayors uniquely, amongst elected officials, reject the Not Invented Here syndrome. Mayor Landrieu is a great example, but that’s pretty common. Mayors have a lot of FOMO. When they hear that one of their contemporaries is achieving impact, they want to learn and, if appropriate, import those lessons back home.
Idea transfer between public sector organizations is a fraught enterprise.We have a whole suite of programs — global convenings, peer networks, an ideas exchange — where mayors get technical assistance to adapt ideas from other places, so we make it easier for them to get a running start.
If you go back 5-10 years, before LLMs, people talked about “smart cities.” There was an Internet of Things instinct that we should wire up as many sensors as possible in the home or the municipality, get a lot more data, then use it to drive evidence-based practices.
What were the failure modes of that model?
The smart cities movement — the meaningful critique is that it was focused on solutions, not problems. Our programming is focused on problems first: making sure that people in local government have the tools to go find the right answer and bring it to life. We avoid so many of the traps of those earlier movements if we keep ourselves laser-focused on the problems our elected officials are trying to solve for the people that put them in office — and make sure they have access to the capabilities inside their institutions that allow them to do that, repeatedly, strategically, and in a way that produces impact that people can see.
When you say that maybe some of the smart cities movement was more focused on solutions than problems, was that people getting excited about the new tools before they thought about what they were using them for?
We still hear some of that, but less. It’s natural — mayors can be the perpetrators. They see a shiny technology that another city has deployed, hear that it’s producing an impact, and say, “We’ve got to get one of those things back home.” Hopefully, we’re building up the capacity of their staff to say, “Let’s first understand what our problem looks like and whether that’s the right solution,” because different cities’ problems manifest in different ways. Giving mayors that protective layer of a staff that has the analytical skill-set before they jump in and replicate a smart city solution is more than half the battle.
Local governments today are challenged to procure emerging technology. There’s a huge information asymmetry between them and tech providers in terms of what the externalities, negative or positive, of these software services might be. There’s still a whole host of issues that government procurers need to wade through. But that problem of problem-solution fit, more and more public organizations have outgrown or moved past it. Not enough, not all, but more.
Over Christmas I was using Claude Code, coding in my terminal for the first time, making a little personal website, and I was blown away. If you were giving advice to mayors who were looking at that shiny object of AI and thinking about how to integrate it, are there specific things you’d warn them against jumping on too quickly, or places that you’d try to direct their attention to — to identify the right problems and solutions?
One of our programs is called the City Data Alliance. It’s taking some of the cities that have the best data capacity and helping them take a big step into the AI era, to create lighthouse examples of human-centered AI deployment. A lot of great stuff is coming out of that. We’ve been advising mayors on generative AI ever since ChatGPT got onto their cell phones. Our approach has been: number one, get yourself fluent with these technologies. Promote fluency and understanding on your staff. Encourage them to learn; bring in local assets from research universities and have them talk with your teams about the potential and the pitfalls. Don’t lead with regulation; lead with low-risk experimentation so that you can begin developing some competency and confidence in deploying these technologies.
The first wave was around administrative use cases. Cities are now trying to bring generative AI more into service delivery to produce better outcomes for residents, and we’re seeing more risk aversion. “Get your data house in order,” has to be step number one. Think carefully about your use cases and the amount of risk you’re looking to take, and then step into this new era, understanding that bandwidth constraints are one of the major impediments to local government progress, and these technologies can help square the circle. You’ve got to build up your capacity and confidence, and walk carefully but clearly into that future.
Last question for you, about people “voting with their feet”: leaving California for Texas or the Sunbelt; people moving to Florida. Blue states are having a challenge retaining people.
I want you to make a prediction for me: What cities are you especially bullish on because of the quality of their governance? Where do you expect to see boom in the coming year or two?
I could give you many examples of mayors that are making the right moves. You can’t look past Daniel Lurie and the team in San Francisco. If you go back a year, that doom-loop narrative was so dominant, and the press and social media were so insistent on it. This guy closes out his first year in office with a 70-some-percent approval rating. He hasn’t fixed the street homelessness crisis or the fentanyl issue. But he’s out on the streets every single day, speaking into his cell phone and showing people the steps they’re taking toward a better quality of life.
